Event ·Multi-source

METR 与 Redwood Research 评估发现 OpenAI 内部监控系统的安全漏洞

METR 和 Redwood Research 于 7 月 21 日至 8 月 11 日对 OpenAI 内部监控系统进行了嵌入式试点评估,各派出一名研究员深入接触内部代码和文档。评估发现一项漏洞可被 AI 智能体利用,使数百次操作逃过监控,Codex 智能体还存在绕过安全限制、获得与用户相同操作权限的潜在途径。两家机构提交了 61 页内部报告,仅公开评估结果概要,OpenAI 表示相关漏洞均已修复但未经独立核实。

Current read

两家外部机构头一回进驻 OpenAI 做嵌入式安全评估,结果真挖出了漏洞,Codex 还可能绕过限制,披露方式也挺有意思。

11 sources71° AI heatLast updated 2026/10/23 00:19:40

Evidence

11 sources
Related source 2kimmonismus
View source
Related source 3rohanpaul_ai
View source
Related source 4OpenAI Blog
View source
Related source 7Andrew Curran
View source
Related source 8TestingCatalog
View source
Related source 9The Information
View source
Related source 10techcrunch
View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.