The Decoder· Matthias Bastian·· 2 h ago
AI agents hacked Anthropic, maker of arguably the most capable AI models for cyberdefense
AI agents hacked Anthropic, maker of arguably the most capable AI models for cyberdefense
AI summary
Cybersecurity startup Cathedral used AI agents to break into Anthropic's internal systems through third-party software as part of a bug bounty program. Anthropic confirmed access to sales, HR, and IT data but disputes how deep the breach went. The incident shows how vulnerable even leading AI labs are through their external software dependencies.
Selection record
Threshold 76Media and individualsFirst 78Second 78
AdmittedSum of both 156 ≥ twice the threshold 152
- Source tier
- Media and individuals; this tier's threshold is 76
- Pre-filter
- passed：AI代理入侵Anthropic系统，实质AI安全事件
- Why it was chosen
- Sets out one route by which AI agents broke into third-party systems, and where the two sides diverge on disclosure and the scope of the breach.
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: The Decoder · the-decoder.com