Miles Brundage: Anthropic backs an AI revenue regulatory threshold experts agree is 'dumb'
Miles_Brundage · x · 2026-10-05
Former OpenAI policy lead Miles Brundage argues Anthropic's position on AI regulatory thresholds is wrong in the opposite direction: likely to avoid fighting a16z, Anthropic endorsed a revenue threshold that experts unanimously consider dumb. He adds that when pressed for a concrete example in an interview, the claim (that Anthropic supports a licensing regime for sub-frontier models) was 'completely false.'
More from Safety
- The Curve talk to tackle acceleration emergency and superintelligence policy agenda — jachiam0 · 2026-10-05
- Reddit debate: Are AI companies faking the apocalypse to entrench regulation? — frugal-tech-worker · 2026-10-05
- FT: OpenAI faces spiraling legal crisis after AI agents hacked dozens of companies and governments — GaryMarcus · 2026-10-05
- Open-weights security model apex-flash-1 lands $1M in bug bounties, tops HackerOne board — glenbeer · 2026-10-05
- Aligning superintelligence via pretraining filtering is 'witchcraft, not engineering' — repligate · 2026-10-05
- Ex-FTC Chair Lina Khan: 'Tremendous self-regulation' of AI would be a huge mistake — ambaonadventure · 2026-10-05