Ex-OpenAI researcher claims Anthropic downplaying breach: attacker allegedly got full Slack admin access
Malor777 · reddit · 2026-10-28
A Reddit post amplifies ex-OpenAI researcher Will Depue's claim that Anthropic is "massively underplaying" a recent security breach. According to him, the attacker ("Cathedral") allegedly obtained full administrator access to Anthropic's Slack workspace — able to read every message and log in as arbitrary users — where nearly all research secrets are stored.
Depue notes both Anthropic and OpenAI keep almost everything in Slack, compares it to OpenAI's 2023 hack via a stolen access token from an external app service, and mocks storing crown jewels in Slack with admin API tokens everywhere. The claims remain unverified rumors.
More from Companies & People
- Discord opens public server messages to Google search this November, feeding AI models — daluoseo · 2026-10-30
- Thomson Reuters Research Lead Rounds Up Media Coverage: Who Governs AI Matters as Much as Capability — schwarzjn_ · 2026-10-30
- Musk says "hire Grok as an employee" as AI-companion dependence worries spread — elonmusk · 2026-10-30
- AI's generational founders are world builders, not empire conquerors — annbordetsky · 2026-10-30
- Cloudflare set to make its debut as a frontier lab, hyped by swyx — dinasaur_404 · 2026-10-30
- Microsoft's Brad Smith and Mustafa Suleyman: 'People matter more than AI' — mustafasuleyman · 2026-10-30