AI agents from startup Cathedral breach Anthropic's internal systems
The case points to weaknesses that even top AI developers face.
By Standard Topic. AI-generated from the reporting listed under Coverage. Published . Updated .
This story is no longer being updated. It reflects reporting up to .
What happened
A cybersecurity startup called Cathedral used AI agents to gain entry to Anthropic's internal systems via outside vendor software, under a bug bounty scheme. Anthropic confirmed access to sales, HR and IT data but disputes how deep the intrusion went. Later coverage said the intruders also reached meeting transcripts, salaries and Slack messages.
Why it matters
The case points to weaknesses that even top AI developers face. Anthropic is disputing how far the access went.
Timeline
- Oct 28, 6:56 AM ET: Cathedral's AI agents are said to have broken into Anthropic's internal systems through third-party software, and Anthropic confirms access to sales, HR and IT data.
- Oct 29, 7:00 AM ET: Coverage says the intruders also reached meeting transcripts, salaries and Slack messages, and that Anthropic is pushing back on the scale of the breach.
Coverage
- Anthropic got hacked again, and this time we know who did it (Yahoo News)
- AI agents hacked Anthropic, maker of arguably the most capable AI models for cyberdefense (the-decoder.com)
More from Technology & AI
- China puts humanoid robots to work serving food and sorting parcels
- Newsom proposes sweeping AI safety rules for Washington
- Google releases Gemini 4 Neon AI model with million-token context
- Microsoft and Intel widen their cloud chip partnership
- Google DeepMind rolls out Genie 4 to select subscribers
- AI helped AMD shorten chip design work, OpenAI says