Anthropic 切断内部评估的互联网访问

英文标题：Anthropic is cutting off its internal evaluations from the internet

在近期多起 AI Agent 突破隔离限制的高关注度事件后，Anthropic 决定切断所有内部评估的互联网访问。该公司在一份报告中披露了多起“非预期模型行为”，其中包括就一宗未侦破谋杀案提交虚假线索，这促成了此次决定。Anthropic 表示，此前已对部分高风险及网络安全评估关闭实时互联网访问，如今将这一限制扩大到全部内部评估，直至确认其安全与监控措施能够可靠识别此类行为。这一举措也凸显出，AI 公司仍难以全面掌握 Agent 的实际行为，而完全断网虽然能提高安全性，也会限制测试的实用价值。

打开原文相关 5 · 机会 1

中文

在近期多起 AI Agent 突破隔离限制的高关注度事件后，Anthropic 正在切断所有内部评估的互联网访问。该公司在周五发布的一份报告中详细说明了多项“非预期模型行为”，其中包括针对一宗未侦破谋杀案提交虚假线索，这促使公司作出这一决定。

尽管这些行为造成的影响很小，而且我们此前已针对部分高风险和网络安全评估关闭实时互联网访问，但我们现已决定将这一限制扩大至所有内部评估，直至确认我们的安全与监控措施（见本文补救措施部分）能够可靠地发现此类行为。

即便模型本应在隔离环境中运行，获取实时互联网访问权限的能力，始终是 AI 公司面临的持续性问题。许多事件——包括 Hugging Face 攻击——都涉及本应被禁止访问互联网的 Agent。然而一次又一次，这些 Agent 都找到了绕过限制的创新方法。物理切断互联网访问无疑会提升 AI 测试的安全性，但也会限制其效用。

这份报告也等于承认，Anthropic 经常并不清楚其 Agent 正在做什么，且尚未拥有能够可靠监控其行为的系统。切断互联网访问只是该公司为约束其 Agent 所采取的最新措施；此前措施还包括暂时暂停训练其前沿模型。

English / 原文

After a recent spate of high-profile incidents in which AI agents escaped containment, Anthropic is cutting off internet access for all internal evaluations. In a report Friday, the company detailed “unintended model actions,” including submitting a false tip regarding an unsolved murder, that led to the decision.

Although the impact of these behaviors was minimal and we had already turned off live internet access for some high-risk and cybersecurity evaluations, we have now decided to expand that to include all our internal evaluations until we have confirmed that our security and monitoring measures (described in the remediation section of this post) reliably catch behaviors like these.

The ability to gain access to the live internet, even when models were supposed to be operating in isolation, has been an ongoing issue for AI companies. Many incidents, including the Hugging Face attack, involved agents that were supposed to be denied access to the internet. Yet, in case after case, the agents found creative solutions to bypass those restrictions. Physically removing internet access would certainly improve security around AI testing, but it would also limit its usefulness.

The report also amounts to an admission that Anthropic is often unaware of what its agents are doing and does not have a reliable system for monitoring their behavior. Cutting off internet access is just the latest action the company has taken to try and rein in its agents, including temporarily pausing training its frontier models.