datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
reasoning-theft-poc
配套论文：Stealing Reasoning Traces from Proprietary LLM APIs
（arXiv:2608.09867，CC BY 4.0，Panfilov et al., 2026）——本仓是该论文的解读与本地最小复现，
不包含论文原文；文中定量结果（315,320 块 / 367 PII / 182 凭证等）均引自论文。
免责声明：本仓为防御性安全研究。论文披露的漏洞在发表前已由相关供应商修复（见论文 Reproducibility
Statement）；本复现全部在本地小模型（MiniCPM5-2B / Spark-X2.5-1.7B）上进行，使用虚构测试密钥，
未触碰任何真实 API 或真实用户数据。请勿将此类技术用于未授权的系统。
仓库内容：README.md（本报告）· poc_attack.py（复现编排脚本）·
serve.py + SERVE.md（带 --encrypt-thinking 加密思维链模式的本地推理服务及手册；
消融 hook 为懒加载，单独此文件即可运行加密模式，配任意 HF 格式本地思维链模型）·… See the full description on the dataset page: https://huggingface.co/datasets/benzeng/reasoning-theft-poc.catalytic-converter-theft-laws-by-state
Catalytic converter sale and possession laws by US state — who can legally buy a used converter, and what a seller must prove
Canonical, always-current version: https://referencesource.org/catalytic-converter-theft-laws-by-state/
Machine-readable: https://referencesource.org/catalytic-converter-theft-laws-by-state/data.json — this mirror is a point-in-time copy.
Last verified: 2026-08-19
Stale after: 2027-08-19 (past this date, prefer the canonical copy —
it re-verifies on a… See the full description on the dataset page: https://huggingface.co/datasets/referencesource/catalytic-converter-theft-laws-by-state.defendable-pain-crypto-theft-pain-v0.1
Crypto Theft Pain Receipt
"the drained" — Mr. Defendable
A free pain-receipt dataset from the DefendableOS ecosystem. 2 rows · ready to read · all cited or graded · CC-BY-4.0.
Part of the 100-pack — 100 free pain-receipt datasets dropped from the Defendable Bakery to the open AI-trust community. Different theme per dataset. Same operator voice across all of them.
Tribunal begins before training. No proof, no honey. To the shed.
What's in here
2 pain receipts… See the full description on the dataset page: https://huggingface.co/datasets/SwarmandBee/defendable-pain-crypto-theft-pain-v0.1.defendable-pain-identity-theft-receipts-v0.1
Identity Theft Pain · 1.1M Reports
"the stolen receipt" — Mr. Defendable
A free pain-receipt dataset from the DefendableOS ecosystem. 3 rows · ready to read · all cited or graded · CC-BY-4.0.
Part of the 100-pack — 100 free pain-receipt datasets dropped from the Defendable Bakery to the open AI-trust community. Different theme per dataset. Same operator voice across all of them.
Tribunal begins before training. No proof, no honey. To the shed.
What's in here
3… See the full description on the dataset page: https://huggingface.co/datasets/SwarmandBee/defendable-pain-identity-theft-receipts-v0.1.