OpenAI Got Hacked By... AI
Summary
Security researchers used Anthropic's Claude to exploit a bug in OpenAI's Discourse forum, gaining access to its private "monorepo" code repository via a bug bounty program. OpenAI also launched Astra for Law, a specialized legal AI tool built on GPT-6 Astra with a dedicated case law index for select firms. Meanwhile, Anthropic redesigned Claude Projects to support parallel agent threads and established a Bay Area wet lab to accelerate AI-driven drug discovery and biological research.
Topics discussed
Intro and daily tech headlines
Sponsor: Vanguard Investor Choice
Security researchers use Claude to hack OpenAI
OpenAI launches Astra for Law
Sponsor: AT&T Connected Car
Sponsor: Home Title Lock
Anthropic redesigns Claude Projects for parallel agents
Anthropic opens Bay Area biology lab
Weekend read: AI and mathematical research
Weekend read: VC resurgence in deep tech
Outro
Listen ad-free on Castria