TREE NEWS update: AI cybersecurity startup Cathedral, founded by former U.S. DOGE members, said its AI agent gained access to Anthropic’s internal systems in under 12 hours, obtaining meeting notes, compensation and hiring information and Slack messages. Anthropic confirmed access to some sales, HR and IT systems, including customer and partner business contact information and some basic user account records, but said Claude user conversations, source code and model weights were not involved. The test ran under Anthropic’s bug bounty program via a third-party commercial software flaw; Anthropic accused Cathedral of incomplete disclosure in violation of program rules.
Cathedral says its AI agent breached Anthropic systems in under 12 hours
Related Coin Charts
AI take
The disclosure fight may matter more than the breach itself: Cathedral frames a sub-12-hour compromise as a demonstration of agentic capability, while Anthropic's objection centers on bug bounty disclosure norms rather than the access obtained. For AI developers and enterprise buyers, the episode suggests agent-driven intrusion is being marketed as a security product, blurring the line between vendor pitch and vulnerability research. Whether such agentic attacks become a standard testing method, and whether bounty programs tighten disclosure rules in response, is the open question.
Generated by AI for reference only.
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.
Related News
3h ago
Texas Brothers Sentenced to 15 Years for Armed Crypto Robbery and Kidnapping
4h ago
Kalshi and Polymarket Diverge on Resolution and Withdrawal Rules
4h ago
Westlake University Launches World Model Research Center
4h ago
Origin Restores OGN Liquidity After Curve Pool Withdrawal
4h ago
CertiK Puts 2026 Crypto Losses at Record $9.2 Billion
4h ago
DeepSeek open-sources Personal Agent architecture