AI Futures team proposes concrete options for pacing the US frontier AI buildout
eli_lifland · x · 2026-09-09
Eli Lifland and colleagues at AI Futures Project published concrete proposals for domestic US regulation to "pace the frontier," responding to an open letter signed by 1,000+ frontier AI employees.
- Pacing means deliberately moderating when AIs above a specified capability level are developed in a jurisdiction, aimed mainly at minimizing existential risk.
- Building on their earlier international proposal "AI 2040: Plan A," they argue domestic regulation may need to come first, then expand into international coordination.
- A side benefit of US-only pacing: slowing what Chinese companies can distill or steal from US models.
- The proposals are explicitly tentative, meant to spark discussion before a more battle-tested version is released.
Related event: AI Futures Project Proposes US Slowdown Plan for Frontier AI(2 posts)→
More from AGI Musings
- Meme Take: The Archons Were the First AI Safetyists — SydSteyerhart · 2026-09-09
- 10,000 agents for 88 hours equal 100 PhD-years of math work — beffjezos · 2026-09-09
- Researcher Who Worked at OpenAI and Anthropic Quits, Accuses Both Labs of Racing to Superintelligence — iamfakhrealam · 2026-09-09
- burkov mocks recurring AGI hype: from counting r's in strawberry to nuclear cheetah fails — burkov · 2026-09-09
- Bronstein: math's authorship chivalry poisoned by a 'cutthroat company' — anshulkundaje · 2026-09-09
- If AI could cause extinction, why don't worried researchers quit? New blog explains — sjgadler · 2026-09-09