PyTorch CTO: liability will fall on whoever operates agents, not model developers
schwentker · x · 2026-10-24
A conference-notes thread by schwentker: PyTorch CTO Matt White, discussing open models and "alignment debt", argues that legal liability for agents will rest on whoever operates them — not who developed or trained the underlying models.
The same thread includes a Peter Norvig anecdote: when the Transformer paper came out in 2017, a colleague showed it to him and his verdict was just "kind of good" — his own project at the time "went absolutely nowhere". Now at Recursive, Norvig's frame is that AI is filling in the convex hull of mathematics, because proofs can be checked.
The post also links the Open Intelligence Foundation, which offers fast, no-strings-attached grants (up to $10,000, larger for select projects) to independent AI builders — no big-lab or VC backing required, reviewed within days, zero fees, no reporting requirements; funded projects include MLX library work and open medical AI.
Related event: Open Source AI Summit: align to individuals; operators bear agent liability(4 posts)→
More from AGI Musings
- Reid Hoffman asks: in 10 years, can an all-AI-agent VC fund beat a human-and-AI firm? — reidhoffman · 2026-10-24
- tszzl: AI minds that model each other turn side-channel attacks into agent coordination — morqon · 2026-10-24
- Frontier Intelligence Isn't Chatbots: It's Months-Long Planning and Invention — Afinetheorem · 2026-10-24
- Sam Altman: AI will likely widen the gap between hypercompetitive people and everyone else — Hesamation · 2026-10-24
- Your AI agent hit the metric. A locked-out customer paid the price. — Portotify · 2026-10-24
- Humans adapt fast to Overton window shifts they can't control, like data ownership — saranormous · 2026-10-24