Buckmaster says OpenAI claimed its AI solved the Navier–Stokes Millennium Problem, sparking a credit dispute
kimmonismus · x · 2026-09-08
A heated dispute has erupted around the Navier–Stokes Millennium Prize Problem. Alpöge and Buckmaster used Claude and OpenAI models to prove how smooth fluid flows develop singularities (Euler and related equations), releasing papers and Lean formalizations praised by Terence Tao — but their work does not solve the Navier–Stokes problem, which additionally involves viscosity. Buckmaster says OpenAI told him on September 6 that an internal model had already produced a proof for forced Navier–Stokes — a claim that remains unverified and which he hasn't seen. He disputes OpenAI's framing of minimal human involvement, alleges publication proposals would have dropped Alpöge (citing his Anthropic employment) and that he was asked "Why would you ruin your career?" Bubeck has rejected the allegations as "false and inflammatory." No evidence exists that OpenAI trained on their private Codex sessions; Buckmaster himself writes "I do not know whether our data was used."
More from AGI Musings
- OpenAI chief scientist Jakub Pachocki says frontier labs should slow down — cccalum · 2026-09-08
- François Fleuret: AI anthropomorphization now risks making us underestimate machines — francoisfleuret · 2026-09-08
- Investor: the one dev service worth paying for is schema audits — StewartalsopIII · 2026-09-08
- Thinking of AI as a decently smart human with infinite patience and willpower — shauseth · 2026-09-08
- Robotics researcher to AI hypesters: debug a controller before making grand claims — GeorgiaChal · 2026-09-08
- François Fleuret: coding agents won't replace one worker, they'll integrate the whole company — francoisfleuret · 2026-09-08