跳到正文
热点事件持续更新

AI agents hacked Anthropic's internal systems

1 篇报道1 个报道来源1 天前更新

先了解这件事

AI 综述

Cybersecurity startup Cathedral used AI agents to breach Anthropic's internal systems through third-party software, accessing sales, HR, and IT data including customer contacts and Slack messages. Anthropic confirmed the breach but stated core assets like Claude chat logs and model weights remained secure. Cathedral exploited vulnerabilities in Anthropic's bug bounty program to demonstrate the attack, which Anthropic says it contained by shutting down affected systems and notifying authorities the same day it was discovered.

AI 根据报道生成 · 1 小时前更新

报道时间线

沿着报道,了解事件的不同侧面。

10月28日
  1. The Decoder精选
    网络安全初创公司利用AI智能体入侵Anthropic内部系统

    网络安全初创公司Cathedral利用AI智能体通过第三方业务软件入侵了Anthropic的内部系统,获取了包括会议记录、薪资数据和Slack消息在内的信息。Anthropic确认部分销售、人力资源和IT系统数据被访问,但表示Claude聊天记录、源代码和模型权重未被泄露。

本事件热度走势

还没有足够的连续观测数据,暂不绘制趋势。