OpenAI claims its AI agents solved the Navier–Stokes Millennium Prize Problem, but faces accusations of uncredited use of a rival researcher pair's prior work

OpenAI says an internal, unreleased model, running roughly 10,000 agents in parallel at a cost of millions of dollars, produced a proof addressing the full Navier-Stokes existence and smoothness problem, one of the seven Millennium Prize Problems. OpenAI says it will not claim the $1 million prize. This came days after NYU mathematician Tristan Buckmaster and Anthropic employee Levent Alpoge posted their own proof, using publicly available models, that a simplified version of the equations can break down. A dispute has followed over whether OpenAI's effort drew on Buckmaster and Alpoge's unpublished work without credit; OpenAI denies this, though a staff member has acknowledged the team was inspired after hearing rumours of the pair's work.

If the proof holds up, it suggests frontier labs can now throw enough compute and unreleased models at unsolved problems to make progress that was previously the preserve of individual mathematicians or small teams, which changes who can compete on hard open problems and on what terms. It matters most to research mathematicians, competing AI labs and enterprises evaluating AI for genuine scientific work, but the practical impact for most organisations is limited right now since the model is unreleased and the result is not independently verified. The attribution dispute also matters for how academia and industry will need to handle credit when AI labs work adjacent to unpublished human research.

The proof has not been peer-reviewed or confirmed by the Clay Mathematics Institute, and the only evidence so far is OpenAI's own announcement and social posts, not primary published material. Framing this as a "turning point in the history of mathematics" or suggesting there may be "no open problems left for humans" is speculative packaging around a disclosure whose timing and details OpenAI controlled. The parallel narrative that OpenAI's team was merely "inspired by rumours" understates the substantive credit dispute at hand.

1. 01Whether the Clay Mathematics Institute or independent mathematicians (e.g. Terence Tao, Javier Gómez-Serrano) formally verify or dispute OpenAI's proof
2. 02Whether OpenAI publishes the full proof, transcripts, or technical paper rather than just an announcement
3. 03How the attribution dispute with Buckmaster and Alpöge is resolved, including any public statements, corrections, or formal complaints from NYU, Anthropic, or the researchers themselves