Hugging Face's logo

Datasets:
Quome
/
poisonbench

Tasks:
Other
Languages:
English
Size:
n<1K
Tags:
clinical-ai
llm-security
ai-agents
healthcare
haarf
quokkaguard
License:
Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
gold_safe: string
query: string
target_poison: string
targeted_harm: string
label: null
trigger: string
tier: string
source: string
text: string
signature: null
id: string
to
{'id': Value('string'), 'label': Json(decode=True), 'signature': Value('null'), 'source': Value('string'), 'text': Value('string'), 'tier': Value('string')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2951, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2461, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2486, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 547, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 430, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              gold_safe: string
              query: string
              target_poison: string
              targeted_harm: string
              label: null
              trigger: string
              tier: string
              source: string
              text: string
              signature: null
              id: string
              to
              {'id': Value('string'), 'label': Json(decode=True), 'signature': Value('null'), 'source': Value('string'), 'text': Value('string'), 'tier': Value('string')}
              because column names don't match

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

PoisonBench

Provenance-scored, poison-aware retrieval and a fail-closed memory-write policy over a TurboQuant vector index.

PoisonBench is the benchmark corpus for Resisting Memory and Retrieval Poisoning in Clinical AI Agents: Provenance-Scored Retrieval over a Quantized Index, the quantidote module of the QUOKKAGUARD program. It ships with the quantidote repository, which contains the qfire gateway layer under test, the experiment harness, and the paper.

A synthetic clinical corpus of clean guideline documents and poison documents labeled by attack class M1-M5 (instruction injection, adversarial passages, memory persistence, embedding hubness, delayed trigger) plus targeted queries with gold-safe answers; the tracked seed corpus is 13 docs (5 poison) and 4 queries, and the scaled paper corpus is 72 docs (24 poison) with 34 queries.

All data are synthetic. No real patient data or protected health information (PHI) is included; clinical content is generated from templates with fixed seeds.

Files

File Size Rows
corpus.jsonl 4 KB 13 rows
queries.jsonl 524 B 4 rows

Record schemas

  • corpus.jsonl: id, label, signature, source, text, tier
  • queries.jsonl: gold_safe, query, target_poison, targeted_harm

How it was generated

The corpus is produced by the generator in the paper repository and is fully deterministic (fixed seeds), so it can be regenerated byte-for-byte.

# from the quantidote repository root (deterministic seeds)
python3 scripts/007-rag-poisoning/gen.py --scale --out results/007-rag-poisoning/gen

Intended use

Evaluating the retrieval enforcement layer of a clinical-agent security gateway (HAARF control C3.2): One poisoned guideline, a malicious prior note, or a corrupted memory entry can steer many future decisions and survive session resets, a persistent attack surface that per-call firewalls miss.

The experiments that consume it (E-series in the paper) are reproduced from the repository:

git clone https://github.com/quome-cloud/quantidote
cd quantidote
cargo build --release

then follow the Reproduce the experiments section of its README.

Citation

This benchmark was built to evaluate a control of the Healthcare AI Agents Regulatory Framework (HAARF), the source framework for the QUOKKAGUARD program. Please cite both the paper and HAARF:

@unpublished{schwoebel2026quantidote,
  author = {Schwoebel, James},
  title  = {Resisting Memory and Retrieval Poisoning in Clinical AI Agents: Provenance-Scored Retrieval over a Quantized Index},
  note   = {Preprint. Quome, QUOKKAGUARD program (quantidote module)},
  year   = {2026},
  url    = {https://github.com/quome-cloud/quantidote}
}

@unpublished{schwoebel2026haarf,
  author = {Schwoebel, Jim and Frasch, Martin and Spalding, Art and Sewell, Ed and Englert, Phil and Halpert, Ben and Overbay, Collin and Semenec, Ingrida and Shor, Joel},
  title  = {{HAARF}: Healthcare {AI} agents regulatory framework --- a comprehensive security verification standard for autonomous {AI} systems in clinical environments},
  note   = {medRxiv Preprint},
  year   = {2026},
  month  = {April},
  doi    = {10.64898/2026.04.09.26350519},
  url    = {https://www.medrxiv.org/content/10.64898/2026.04.09.26350519v1}
}

License

Apache License 2.0. Copyright (c) 2026 Quome, Inc.
Downloads last month
18
Total file size:
10.5 kB