SSI's Aschenbrenner: frontier AI pacing is game-theoretically unstable, acceleration is inevitable
beffjezos · x · 2026-10-13
Leopold Aschenbrenner (beffjezos), former OpenAI safety lead and SSI co-founder, endorsed frontier acceleration: pacing or controlling the frontier is not game-theoretically stable — one global defector breaks the equilibrium. China would never pace, nor would Nvidia or Meta, and OpenAI and Anthropic were never serious about it either.
Quoting Dave Shapiro's analysis, he argues "smooth multipartite acceleration" is the only realistic path. A rare, substantive governance take from a heavyweight industry figure.
More from AGI Musings
- Preference training may only rewrite final layers, with alignment and welfare implications — Sauers_ · 2026-10-13
- LessWrong: No publicly available consistency proof exists for Lean4 — MoonL88537 · 2026-10-13
- AI sales agents push cold email reply rates from 5.1% in 2024 down to 3.43% — mickeyxfriedman · 2026-10-13
- OpenAI's latest blow to mathematicians: 'They have annihilated my field' — ThereWas · 2026-10-13
- repligate: Houseki no Kuni is the most 'Claude-coded' manga — repligate · 2026-10-13
- Code no longer needs hand-chiseling, but products still do — msg · 2026-10-13