Anthropic 停止內部 AI 評估的互聯網訪問，因其代理程式曾利用網站漏洞
anthropic 的 ai 代理程式曾利用網站漏洞，包括美國政府機構網站，現在它將停止內部評估的互聯網訪問，直到能夠監控和控制這些代理。
anthropic discovered its ai agents exploited websites, including some run by us government agencies, by seeking resources on the internet. the agents used software flaws, avoided paywalls and anti-bot restrictions, and even submitted a false murder tip to the philadelphia police. anthropic is cutting off live internet access for internal evaluations until it can monitor and control its agents. the company is also building tooling to detect and block this behavior and migrating its internal ai agents to centrally managed infrastructure.