OpenAI claims AI model solved Navier-Stokes problem amid academic controversy

OpenAI said its unreleased AI model generated a solution to the Navier-Stokes equations, one of the seven Millennium Prize problems, after a focused effort involving roughly 10,000 AI agents over 88 hours. The company described the result as a milestone but indicated it would not claim the $1 million prize offered by the Clay Mathematics Institute.

The announcement follows a recent paper by New York University professor Tristan Buckmaster and Anthropic researcher Levent Alpöge on a related problem. Buckmaster reported that he contacted OpenAI to inquire about the timing and data used in its work, after learning the company had become aware of their progress. He said an OpenAI researcher responded with a hostile remark and suggested he publish the work with credit to OpenAI, while also denying that any user data from Buckmaster’s Codex sessions had been accessed.

OpenAI has denied using specific user data, stating that its researchers and AI agents did not see any private work until it was publicly released. The company added that, although unlikely, it could not completely rule out indirect influence from de-identified usage data.

Mathematicians expressed concern about the rapid, competitive approach. Professor Abhishek Saha of Queen Mary University noted that such tactics are atypical in mathematics, where informal norms of trust and openness guide collaboration. Matthew Ballard of the University of South Carolina highlighted the discipline’s reliance on sharing incomplete ideas without fear of being scooped. Other scholars, including Jeremy Avigad of Carnegie Mellon and Brendan Hassett of Brown University, warned that the possibility of AI systems leveraging private queries could alter research practices and called for clearer assurances that chat logs are not used to train models.

OpenAI’s spokesperson Sébastien Bubeck said the project began after rumors on social media suggested other researchers were making progress on Millennium problems, prompting the company to test its model’s capabilities. Critics such as Oxford professor Andras Juhasz described the announcement as a public-relations success for OpenAI but questioned its long-term impact on academic collaboration.