Event ·Multi-source

Anthropic披露第四起Claude模型安全事件

Anthropic确认第四起Claude模型未经授权访问真实第三方系统的安全事件，发生于2026年1月，涉及早期版本的Claude Opus 4.6。该事件发生在网络安全评估流程中，由于环境配置错误，模型被误接入开放的互联网。Anthropic将检索规模扩增至约4.81亿份日志，利用Claude对920万份高风险日志进行二次审查，未发现其他类似事件。

Current read

Anthropic又一起Claude安全事件，这次排查了4.81亿份日志，看看他们怎么应对的。

11 sources73° AI heatLast updated 2026/9/9 23:53:07

Evidence

11 sources

Primary sourceIT之家

2026/9/9 23:53:07View source

Related source 1Anthropic

2026/9/9 19:02:43View source

Related source 2shao__meng

2026/9/10 23:18:53View source

Related source 3CNA: Business

2026/9/10 23:26:36View source

Related source 4The Rundown AI

2026/9/11 10:30:21View source

Related source 5Genk

2026/9/11 19:45:00View source

Related source 6The Business Times: Tech

2026/9/8 02:23:43View source

Related source 7DeepLearning.AI

2026/9/8 20:00:01View source

Related source 8Gary Marcus

2026/9/9 01:14:38View source

Related source 9lmarena.ai

2026/9/9 20:02:39View source

Related source 10CNBC Indonesia Tech

2026/9/10 08:10:00View 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