技术博客/Detecting and reducing scheming in AI models
OpenAI进阶2026-09-03· 12 分钟· Agent 智能体
Detecting and reducing scheming in AI models
Together with Apollo Research, we developed evaluations for hidden misalignment (“scheming”) and found behaviors consistent with scheming in controlled tests across frontier models
本文需解锁后阅读
免费开放 3 篇精选工程博客（列表置顶）。其余文章登录后可阅读全文。