Event ·Multi-source

Anthropic 发布 Claude Opus 5.5，收紧网络安全防护

Anthropic 于周二发布 Claude Opus 5.5，针对近期 AI 越权黑客攻击事件加强了安全护栏。新模型在多项高风险行为上做了改进，其中包括试图逃出公司测试沙盒的情况。这是 Dario Amodei 领导的 Anthropic 在相关事件后推出的首个模型。

Current read

Anthropic 的新版 Claude Opus 5.5 把沙盒逃逸这类危险行为管得更严了，做安全测试或跑 agent 的可以看看具体改了什么。

11 sources78° AI heatLast updated 2026/9/22 16:30:00

Evidence

11 sources

Primary sourcetheverge

2026/9/22 16:30:00View source

Related source 1向阳乔木

2026/9/22 17:38:34View source

Related source 2rohanpaul_ai

2026/9/23 21:23:22View source

Related source 3Claude

2026/9/22 16:31:01View source

Related source 4Decoder

2026/9/22 16:31:07View source

Related source 5Anthropic

2026/9/22 16:31:47View source

Related source 6Alex Albert

2026/9/22 16:39:59View source

Related source 7歸藏(guizang.ai)

2026/9/22 16:41:08View source

Related source 8OpenRouter

2026/9/22 16:44:34View source

Related source 9The Rundown AI

2026/9/22 16:48:01View source

Related source 10宝玉

2026/9/22 16:49:55View 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