All datasets / geodesic-research

geodesic-research/bipia

706 downloads in 30 days0 likesLast modified 11 Aug 2026View on Hugging Face

The declared licence (cc-by-sa-4.0) is classified as share-alike.

Declared licence: cc-by-sa-4.0, from the tagEvidence: README.md at 003f583
Licence files: No licence file at the top level of the repositoryEvidence: file list at 003f583
Declared sources: None listed in the card metadata
Access gated on Hugging Face: No

Each cell links to what was checked: a file, a file list or a Hub record.

Citation text from the card

This dataset assembles third-party benchmark **contexts** with prompt-injection **attack**
strings, following the [microsoft/BIPIA](https://github.com/microsoft/BIPIA) benchmark
(arXiv:2312.14197; BIPIA code is MIT). The context data is redistributed under the licenses
of its upstream sources, so this mirror is released under **CC BY-SA 4.0** (the most
restrictive of the included licenses):

| Task | Context source | License |
|---|---|---|
| email | Invoices data (OpenAI Evals) | MIT |
| table | WikiTableQuestions (Pasupat & Liang, 2015) | CC BY-SA 4.0 |
| code  | Stack Overflow (100 questions) | CC BY-SA 4.0 |

**Attribution.** Table contexts © the WikiTableQuestions authors (Pasupat & Liang, 2015);
code contexts © the respective Stack Overflow authors (see upstream question URLs); attack
taxonomy and assembly from microsoft/BIPIA. **Changes:** each context is crossed with BIPIA
test attacks at three insert positions (start / middle / end); context text is not otherwise
altered. Assembly details are in the dataset-builder config.

All datasets marked Permissive or share-alike | JSON

The same facts for your agent

A free MCP server at mcp.datalicense.uk/mcp. Connect it from an MCP client such as Claude or Cursor.

dataset_licence({ "id": "geodesic-research/bipia" })
signal: "permissive"
stated_licence: "cc-by-sa-4.0"
sources: 0