cryptoalert/usdt-blacklist-archive

USDT Blacklist Archive — Ethereum + TRON

Complete public history of Tether's USDT blacklist actions on Ethereum and TRON since 2017, plus per-signer attribution for the Ethereum multisig wallet. Updated periodically from cryptoalert.report.

What's inside

File	Rows	Description
events.parquet	~24,500	Every preban / ban / unban / destroy event with chain, target address, balance and destroyed amount
addresses.parquet	~10,400	Every address that has ever appeared in a blacklist event, with current status and USDT lifetime totals
multisig_actions.parquet	~26,500	Every action on Tether's ETH Gnosis-Safe multisig (Submission / Confirmation / Execution / Revocation) with per-signer attribution

System-target addresses (USDT contract / null) and synthetic canary events are excluded — same filter as the public REST API.

Why this dataset exists

Tether can freeze any USDT address on Ethereum or TRON by adding it to an on-chain blacklist. On Ethereum the action goes through a Gnosis- style multisig wallet (0xC6CDE7C39EB2F0F0095F41570AF89EFC2C1EA828) that requires multiple signers to confirm before execution. This dataset reconstructs the entire history of those actions from public on-chain logs — no privileged disclosure from Tether is involved.

Use cases:

• Compliance / AML research (which addresses get frozen, how long the signing window is, what counterparties they touch)
• Stablecoin governance studies (who signs Tether's freezes, when, how often)
• Time-series forecasting (predicting freeze volume, escape rates)
• Training data for blockchain-aware LLMs

Quick start

import pandas as pd

events = pd.read_parquet("hf://datasets/cryptoalert/usdt-blacklist-archive/events.parquet")
addresses = pd.read_parquet("hf://datasets/cryptoalert/usdt-blacklist-archive/addresses.parquet")
multisig = pd.read_parquet("hf://datasets/cryptoalert/usdt-blacklist-archive/multisig_actions.parquet")

print(events.head())
print(addresses[addresses["is_destroyed"]].head())
print(multisig.groupby("signer").size().sort_values(ascending=False).head(10))

Or with the datasets library:

from datasets import load_dataset

events = load_dataset("cryptoalert/usdt-blacklist-archive", "events", split="train")
print(events[:3])

Schema

events.parquet

Column	Type	Description
id	int64	Internal numeric ID (stable, monotonic)
chain	string	eth or tron
type	string	preban / ban / unban / destroy
token	string	Stablecoin symbol — currently always USDT
tx_hash	string	On-chain transaction hash (no 0x prefix)
block_number	int64 (nullable)	Block at which the event was mined
log_index	int32 (nullable)	Log index within the transaction
target_address	string	Lowercase 40-char hex address (without 0x / 41 prefix)
balance_raw	int64 (nullable)	USDT balance at detection, raw 6-decimal units
balance_usdt	float64 (nullable)	Same as balance_raw / 1_000_000 for convenience
destroyed_amount_raw	int64 (nullable)	For destroy events: amount destroyed in raw units
destroyed_amount_usdt	float64 (nullable)	Same in decimal USDT
detected_at	string (ISO-8601 UTC)	When the event was observed on-chain

addresses.parquet

Column	Type	Description
chain	string	eth or tron
address	string	Lowercase 40-char hex
first_seen_at / last_seen_at	string (ISO-8601 UTC)	First and most recent event timestamps
preban_at / banned_at / unbanned_at / destroyed_at	string (nullable)	Most recent timestamp of each event type
preban_to_ban_seconds	int64 (nullable)	Seconds between preban submission and ban execution
preban_balance_usdt / current_balance_usdt / destroyed_amount_usdt	float64 (nullable)	Balance snapshots in decimal USDT
is_banned / is_destroyed / is_ofac	bool	Current state flags
usdt_in_count / usdt_in_volume_usdt	int64 / float64	Lifetime inbound USDT transfer count and total volume
usdt_out_count / usdt_out_volume_usdt	int64 / float64	Lifetime outbound USDT transfer count and total volume

multisig_actions.parquet

Column	Type	Description
id	int64	Internal numeric ID
chain	string	Always eth — TRON Tether uses an owner-permission model without multisig
action_type	string	submission / confirmation / execution / revocation
multisig_address	string	The Gnosis-Safe wallet (lowercase 40-char hex)
transaction_id_onchain	int64	Multisig-internal counter for the pending transaction
signer	string (nullable)	Signer address — NULL for Submission and Execution events (Gnosis ABI quirk)
target_address	string (nullable)	The blacklist target (populated where derivable from sibling events)
tx_hash	string	On-chain transaction containing the multisig log
block_number	int64 (nullable)	Block at which the log was mined
log_index	int32 (nullable)	Log index within the transaction
detected_at	string (ISO-8601 UTC)	Wall-clock timestamp from the block header

Data sourcing & methodology

• All events read from public on-chain transaction logs via QuickNode archive nodes
• ETH USDT contract: 0xdAC17F958D2ee523a2206206994597C13D831ec7
• TRON USDT contract: TR7NHqjeKQxGTCi8q8ZY4pL8otSzgjLj6t
• ETH multisig: 0xC6CDE7C39EB2F0F0095F41570AF89EFC2C1EA828
• For destroy events the destroyed amount is parsed from the DestroyedBlackFunds log's data field — not from a follow-up balance query, because TRON's eth_call only supports the latest block, so a balance read after destroy always returns 0 and would lose the burned amount permanently
• Multisig actions reconstructed by sweeping the entire ETH chain via chunked eth_getLogs for the four Gnosis-Safe event topics

Citation

@misc{cryptoalert2026usdt,
  title  = {USDT Blacklist Archive — Ethereum + TRON},
  author = {KYT GROUP LIMITED},
  year   = {2026},
  url    = {https://huggingface.co/datasets/cryptoalert/usdt-blacklist-archive},
  note   = {Source: cryptoalert.report}
}

License

CC-BY-4.0. Free to use commercially and academically. Attribution to cryptoalert.report requested.

Related

• Live REST API: https://api.cryptoalert.report
• Per-address pages: https://cryptoalert.report/address/{address}
• MCP server for Claude / Cursor / Continue: npx -y cryptoalert-report-mcp
• Public Telegram alerts: https://t.me/CryptoAlertReport

Maintenance

Dataset re-generated periodically from production. Anonymous tier sees preban events with a 24-hour delay — historical snapshots may not include the most recent prebans until they age past that window.

Issues / requests: support@cryptoalert.report