Dario Amodei's 'We Must Pace the Frontier': Anthropic commits to permanent third-party eval access
Miles_Brundage · x · 2026-09-13
Anthropic CEO Dario Amodei published a new essay, "We Must Pace the Frontier," arguing the AI industry should slow down and laying out a three-part plan. As a first step, Anthropic is unilaterally committing to give third-party evaluators permanent, employee-level access to its systems — to verify safety measure adherence, report incidents, and assess alignment during training. Commentators including former OpenAI policy lead Miles Brundage highlighted the need for agreements or regulation (ideally both), especially for open-weight Chinese models, with push needed from the White House and Congress; verification emerged as the critical open problem.
Related event: Anthropic CEO Dario Amodei Calls on Industry to Pace the Frontier(84 posts)→
More from AGI Musings
- Timnit Gebru: AI companies stoke extinction fears to dodge real harms like autonomous weapons — WellsLucasSanto · 2026-09-13
- Ethan Mollick: orgs underestimate AI progress, the AI world underestimates its jaggedness — emollick · 2026-09-13
- 33-author paper maps the road to recursive self-improvement, reviewing 72 AI teams — huybery · 2026-09-13
- AI doomers get called out: moral signaling without building isn't action — nptacek · 2026-09-13
- 33-author arXiv paper lays out a roadmap toward genuine recursive self-improvement — chaumian · 2026-09-13
- Open-source camp on lab pacing: the world won't accept superintelligence owned by two companies — willcb · 2026-09-13