OpenAI claims Navier-Stokes breakthrough amid plagiarism dispute

September 22, 2026 at 08:23 PM2 min

This digest was compiled by AI from multiple sources — links to the originals are below.

OpenAI announced a solution to the Navier-Stokes Millennium Prize problem on September 5. NYU mathematician Tristan Buckmaster says the company may have used an approach he and Levent Alpoge developed over nearly a year. OpenAI denies the claim and says it will not seek the $1 million prize.

Key Facts

OpenAI's model began working on the Navier-Stokes problem on September 1 and produced a full proof by September 5.
Tristan Buckmaster and Levent Alpoge found a key approach to the problem in August, before OpenAI's announcement.
OpenAI used 10,000 AI agents running for 88 hours to produce the proof.
Three mathematicians showed that OpenAI's approach fails when the external force is removed, according to Scientific American.
OpenAI will not claim the $1 million Clay Institute prize.

The Plagiarism Dispute

Tristan Buckmaster, a mathematician at New York University, said OpenAI may have used an approach he developed with Levent Alpoge over nearly a year. Buckmaster and Alpoge found an important path to solving the Navier-Stokes problem in August. In early September, Buckmaster learned that OpenAI's experimental model had also reached a solution using what he described as the same unusual approach. Buckmaster said he has not seen OpenAI's proof and does not know whether the company used his materials.

The Mathematical Dispute

OpenAI's system considered a version of the problem in which an external force acts on the fluid. This version is included in the official Clay Institute formulation, but many mathematicians consider the main question to be the behavior without such a force. Three mathematicians recently showed that OpenAI's approach cannot be transferred to the no-force version, according to Scientific American. Their calculations indicate that removing the external force eliminates the infinite-velocity problem described in the proof.

Publication Negotiations

Buckmaster said he was offered a joint publication without Alpoge because Alpoge works at competing company Anthropic. OpenAI's head of mathematics, Sebastian Bubeck, confirmed that such an option was discussed but said it concerned a separate publication, not excluding Alpoge from his own work. The proof must undergo independent verification, and the Clay Institute must separately decide whether it meets the requirements for the problem to be considered solved.