OpenAI Built Navier-Stokes Proof on Buckmaster's Work — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Tristan Buckmaster and Anthropic mathematician Levent Alpöge announced three proofs on the Navier-Stokes existence and smoothness problem — a $1 million Millennium Prize problem — using Codex and Claude models.
- OpenAI published a full proof days later using an unreleased next-generation model, with the week-long effort consuming 300 billion output tokens — $22.5 million in compute at current Astra rates.
- Buckmaster alleges information about his team's progress reached OpenAI before the September 1 start of its effort, and that the identical uncommon approach (smooth force, options c and d in Fefferman's statement) was 'not the direction one arrives at in a few days by giving a model the problem statement.'
- OpenAI mathematician Sébastien Bubeck allegedly asked Buckmaster to remove Alpöge's credit as a compromise, and when Buckmaster pushed to go public, said 'Why would you ruin your career?' followed by 'If you don't want me to be nice, then I don't have to be nice.'
- Buckmaster flagged that his extensive Codex use could have leaked his approach, since OpenAI reserves the right to train on Codex interactions — a concern OpenAI downplayed but did not fully rule out, citing only that proofs 'differ significantly.'
- OpenAI confirmed the September 1 timeline and ongoing conversations with Buckmaster, while maintaining its team 'did not see any of their work through any means until they released it publicly.'
Why it matters: The dispute exposes a conflict Buckmaster himself flagged: OpenAI reserves the right to train on Codex interactions while racing those same users to publish results, turning the platform itself into a vector for competitive intelligence. Bubeck's alleged 'ruin your career' warning shows how directly corporate AI labs now collide with academic priority norms, with $22.5 million in compute behind the collision.
Ask SkimNews
