METR 与 Redwood Research 员工进驻 OpenAI 评估内部智能体监控
Why we picked it
METR 和 Redwood Research 的人直接进 OpenAI 办公三周，专门审它监控内部智能体的系统，报告已经公开，做 AI 安全评估的可以看看。
METR 和 Redwood Research 的两名员工今年夏天在 OpenAI 驻场工作三周，参与一项嵌入式评估试点。团队评估的是 OpenAI 用来监控内部智能体对齐状况的系统。METR 已发布这项工作的简要报告。
Original · METR
Two employees from METR and Redwood Research worked on premises at OpenAI for three weeks this summer in an embedded assessment pilot. This team evaluated the systems OpenAI was using to monitor for misalignment from internal agents. A brief write-up of their work is now public.
