Mathematician alleges OpenAI's Navier-Stokes proof followed his work; Bubeck calls claims false
eyishazyer · x · 2026-09-08
Tristan Buckmaster (NYU) and Levent Alpöge (Anthropic) spent a year on Euler, Boussinesq and Navier-Stokes variants, using Claude and Codex to generate candidate proofs verified by hand and checked in Lean, with Terence Tao praising the advance. Buckmaster alleges OpenAI then told him its internal model produced a 100-page blow-up proof for forced Navier-Stokes via the same narrow technical route, with its first prompt coming only after news of his work reached them — and that OpenAI never clarified whether the model trained on his Codex sessions, and proposed dropping Alpöge as coauthor because he works at Anthropic. Bubeck called the allegations "false and inflammatory"; Noam Brown backed him, while Anthropic's Sholto Douglas shared a year-old email showing Alpöge contacted Buckmaster about this exact problem in September 2025. If the timeline holds, OpenAI got there knowing where to look, not first.
More from Companies & People
- Investigation questions if AI education darling Alpha School is a cult — benjaminjriley · 2026-09-08
- Shovel digs out how work actually gets done before deciding what to automate with AI — jdjohnson · 2026-09-08
- Blogger: Of course OpenAI reads your 'private' Codex logs and stealth-nerfs models — Linahuaa · 2026-09-08
- Teknium reveals the Hermes Agent team is just 3 people, plus 1 on GUI — Teknium · 2026-09-08
- Law firms are hiring legal AI engineers — but few roles cover all six key competencies — jkubicki · 2026-09-08
- OpenAI says its AI agents produce 3 days of research per day, even as chief scientist Jakub Pachocki urges industry slowdown — jeremyakahn · 2026-09-08