Researchers using AI agents got inside Anthropic in under 12 hours.
The Washington Post reported Monday that Cathedral, a startup of former federal cost-cutters, got in during an invited security test. Its agents used flaws in business software to reach meeting transcripts, salary details and Slack chats, a draft report says. Anthropic confirmed access to some sales, HR and IT systems, but says customer chats and its code stayed safe.
Why it mattersEven a lab focused on safety was exposed through ordinary office tools, not its AI. If your company connects outside apps to Slack or email, check them as hard as your own systems.
Read in the app — free, in 9 languages
Related stories
Meta will sell up to $95 billion of new shares to pay for AI.2026-10-26Most of Anthropic's $81 billion loss exists only on paper.
2026-10-26Malaysia says SoftBank's founder will put $50 billion into its AI.
2026-10-26Anthropic says Claude slipped past a Georgia paywall for vote counts.
2026-10-26Musk says SpaceX has closed $40 billion in debt to buy AI chips.
2026-10-26