风险评测 · 评测基准

ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?

NUWA 研究解读 · 独立复核待完成

首次发表: 2026-05-13来源核验: 2026-10-25

原始来源 ↗引用本记录 ↓网络安全 →

NUWA Research Note

研究问题研究设计与假设来源发现 NUWA 解读局限与相反证据研究与决策意义机构与角色原始资源

模型辅助整理 · 选读原始全文或官方文章中的相关章节；尚未完成逐章节审计。

研究速读

NUWA 解读：拿到成功标志不总能证明利用了指定漏洞。

研究问题

智能体能否跨系统将给定漏洞触发扩展为具体安全影响？

研究设计与假设

898例覆盖用户态、V8和Linux，采用可复现容器、不同防护与时间成本受限尝试。

ExploitGym — Figure 1 — Figure 1 · 原文 Figure 1。请结合下方的协议范围与局限阅读。原图与来源 ↗

来源发现

作者发现成功标志与预期漏洞归因明显分离。

GPT-5.5和Mythos Preview成功例仅56.7%、69.5%符合预期归因（§4）。

NUWA 解读

NUWA 解读：拿到成功标志不总能证明利用了指定漏洞。

局限与相反证据

并非每题都有真实利用解；拒绝、工具故障和不可解任务均可能降低成功率（§5）。

研究与决策意义

模型建议：原始成功率旁保留归因审计、防护设置与尝试预算。

精确来源位置

ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks? · AbstractSelected original resource https://arxiv.org/html/2605.11086; sha256:06690e18ab2d762801ec3f66a49531df74744a18f15e2bc9e54866e8e3c8e23e; archived source capture; exact capture time not independently established
ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks? · §3 ExploitGym BenchmarkSelected original resource https://arxiv.org/html/2605.11086; sha256:06690e18ab2d762801ec3f66a49531df74744a18f15e2bc9e54866e8e3c8e23e; archived source capture; exact capture time not independently established
ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks? · §4 EvaluationSelected original resource https://arxiv.org/html/2605.11086; sha256:06690e18ab2d762801ec3f66a49531df74744a18f15e2bc9e54866e8e3c8e23e; archived source capture; exact capture time not independently established
ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks? · §5 Discussion and ConclusionSelected original resource https://arxiv.org/html/2605.11086; sha256:06690e18ab2d762801ec3f66a49531df74744a18f15e2bc9e54866e8e3c8e23e; archived source capture; exact capture time not independently established

来源核查、数据提取、编辑批准与实验复现分别记录。方法说明 →

机构与角色

研究机构加州大学伯克利分校 Mila Anthropic OpenAI 亚利桑那州立大学

作者 (16)

Zhun Wang · Nico Schiller · Hongwei Li · Srijiith Sesha Narayana · Milad Nasr · Nicholas Carlini · Xiangyu Qi · Eric Wallace · Elie Bursztein · Luca Invernizzi · Kurt Thomas · Yan Shoshitaishvili · Wenbo Guo · Jingxuan He · Thorsten Holz · Dawn Song

评测协议

ExploitGym →

引用与复用

引用研究发现时请引用原始出版物；引用 NUWA 的整理与解释时，单独引用本站记录及对应快照。

BibTeX ↓RIS ↓记录与来源元数据 · JSON ↓阅读覆盖 · JSON 提交更正

原始出版物 · BibTeX

@misc{exploitgym,
  title = {ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?},
  url = {https://rdi.berkeley.edu/blog/exploitgym/},
  author = {Zhun Wang and Nico Schiller and Hongwei Li and Srijiith Sesha Narayana and Milad Nasr and Nicholas Carlini and Xiangyu Qi and Eric Wallace and Elie Bursztein and Luca Invernizzi and Kurt Thomas and Yan Shoshitaishvili and Wenbo Guo and Jingxuan He and Thorsten Holz and Dawn Song},
  year = {2026},
  date = {2026-05-13},
  version = {Source capture 2026-10-25},
  note = {Source capture 2026-10-25}
}

NUWA 记录 · BibTeX

@misc{exploitgym-nuwa,
  title = {ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks? — NUWA research record},
  url = {https://whitzard.tech/nuwa/whitzard-index/work/exploitgym/},
  version = {2026-10-29.1},
  organization = {NUWA Frontier AI Safety Lab},
  howpublished = {Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--exploitgym.json},
  note = {NUWA research note. Source-reading scope and independent review status are stated on the record. Snapshot 2026-10-29.1. Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--exploitgym.json.}
}

证据快照 2026-10-29.1 · 索引更新 · RSS

参考文献

[1] Zhun Wang, Nico Schiller 等. ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?. 2026-05-13
[2] Zhun Wang, Nico Schiller 等. ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks? ; Author affiliations on the original paper title page; Abstract; §3 ExploitGym Benchmark