Event ·Multi-source

METR 评审 Anthropic 2026 年 8 月风险报告中的自动化研发风险章节

METR 发布了对 Anthropic 2026 年 8 月版 Risk Report 中'Risks from automated R&D'章节的评审意见，评审过程中还获得了非公开版本的内容。评审重点集中在两处：该报告关于自动化 AI 研发可能带来灾难性风险的论断，以及 Anthropic 是否已跨过相关能力阈值。这一评审针对的是 Anthropic 自行发布的企业风险报告，而非政府监管文件。

Current read

METR 拿到了 Anthropic 风险报告的非公开版本，专门评了自动化 AI 研发那部分，关心 AI 安全的可以看看它怎么挑刺。

11 sources61° AI heatLast updated 2026/11/5 18:58:49

Evidence

11 sources

Primary sourceMETR

2026/11/5 18:58:49View source

Related source 1Anthropic

2026/11/3 20:10:08View source

Related source 2arXiv: DeepSeek

2026/11/3 19:39:03View source

Related source 3rohanpaul_ai

2026/11/3 21:07:00View source

Related source 4Andrew Curran

2026/11/3 22:02:31View source

Related source 5marktechpost

2026/11/3 22:08:22View source

Related source 6techcrunch

2026/11/3 22:41:03View source

Related source 7IT之家

2026/11/4 00:00:32View source

Related source 8AIGCLINK

2026/11/4 01:08:31View source

Related source 9VnExpress Số hóa

2026/11/4 08:00:00View source

Related source 10Decoder

2026/11/4 15:23:49View 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.

Open the story