Cathedral's AI agents breached Anthropic's third-party systems, and the scope is disputed
AICybersecurity startup Cathedral used AI agents to access some Anthropic internal sales, HR, and IT systems through third-party business software, according to a draft report reviewed by The Washington Post. Anthropic says the researchers reached only limited data, including some customer and partner contact details and basic user account data, and that Claude chats, source code, and model weights were not exposed.
AIWhy it matters.
The source separates Anthropic's account of the breach from Cathedral's and a critic's claims, showing how much of the incident's scope remains disputed.
Source details
The Decoder · Matthias Bastian · the-decoder.comPublished
Original title: AI agents hacked Anthropic, maker of arguably the most capable AI models for cyberdefense