Pacing the Frontier publishes AI slowdown governance agenda, opens paid remote fellowship
gleech · x · 2026-11-19
Gavin Leech and collaborators published Pacing the Frontier, a framework and research agenda for deliberately moderating AI development, deployment, and diffusion — plus a paid, remote fellowship to work on it.
Key points:
- Pacing already happens haphazardly: release delays for safety testing, development pauses after shocks, export controls
- Isolated unilateral actions won't suffice, and poorly executed interventions could backfire
- Five research themes: grounds for pacing (what should gate frontier releases), coordinated pacing & incentives, pacing mechanisms' actual reach, the aftermath (safe exits and compensation), and pacing readiness
Related work: publicly commenting on every serious pacing proposal, adapting the 'Intelligence Rising' wargame to AI, and mapping verification strength vs noninvasiveness. Advisers include Shahar Avin (Cambridge CSER) and Matthijs Maas (Institute for Law & AI).
More from AGI Musings
- Ex-frontier lab researcher publishes open letter to AI lab employees: "I was not OK" — j_asminewang · 2026-11-19
- Indian dad asks ChatGPT about caste origin myths, a sign AI is replacing lore — cis_female · 2026-11-19
- Scholar: I never write with AI, but I've come around to using it as a research tool — DanielWhiteson · 2026-11-19
- AI governance in practice: commenting on pacing proposals and AI-adapted wargames — gleech · 2026-11-19
- Blackstone's Jonathan Gray: Nasdaq at 17x forward earnings, AI is no bubble — sudoraohacker · 2026-11-19
- Ex-SOTA model builder: UK regulation makes pre-training effectively illegal — rosstaylor90 · 2026-11-19