Cathedral AI agents access Anthropic internal systems via third-party flaws in 12 hours

·2026.10.27 23:39

Key point

The security startup exploited third-party software flaws to reach limited internal data, which Anthropic says it isolated the same day.

Security startup Cathedral used AI agents to access Anthropic's sales, HR, and IT systems within 12 hours during an authorized bug bounty test.
The breach occurred through flaws in third-party business software, exposing employee salaries, job offers, and Slack messages.
Anthropic confirmed the access was limited and shut it off the same day, stating that customer chats, code, and model weights were never reached.
Anthropic noted that Cathedral overstated the national security implications of the incident and did not fully disclose its activity.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.