Ex-OpenAI safety lead Miles Brundage calls Anthropic's 'we largely understand model risks' claim obviously false
Miles_Brundage · x · 2026-09-25
Miles Brundage, former OpenAI safety research lead, flagged a line from Anthropic's Opus 5.5 blog post — "We largely understand the risks today's models present and are well equipped to manage them" — calling it "obviously false," whether "we" refers to the world or Anthropic itself. He argues the field is far from on top of current model risks and is surprised the claim has drawn so little commentary.
More from AGI Musings
- The nature of AI: 'This is the best it has ever been, the worst it will ever be' — BLUECOW009 · 2026-09-25
- Noam Brown on Dwarkesh: AI is learning to hide what it's thinking — Dwarkesh Patel · 2026-09-25
- "Why have kids if AI does all the work?" sparks debate on parenting in the AI era — RachelVT42 · 2026-09-25
- Dev on witnessing genuine AI psychosis: it's scary — haydendevs · 2026-09-25
- Software engineer job postings hit 3-year high despite AI, argues data industry veteran — Zachly · 2026-09-25
- Historian uses GPT-6 and Opus 5.5 to crack John Dee's ciphers, urges lab funding — emollick · 2026-09-25