OpenAI announced on September 8, 2026, that an unreleased internal AI model solved the historic Navier-Stokes equations, a Millennium Prize Problem. The claim immediately triggered a fierce credit dispute with independent researchers who alleged the company built on their unreleased work.
The Million-Dollar Problem and the OpenAI Breakthrough
The Navier-Stokes equations, which describe how fluids like air and water move, have baffled mathematicians for generations. Classified in 2000 by the Clay Mathematics Institute as one of seven Millennium Prize Problems
carrying a $1 million reward, the equations are used in everything from weather forecasting and aircraft design to the study of blood flow. Until now, only one of the seven problems had been solved.
OpenAI stated that its unreleased model solved the challenge by showing that these equations can break down over time. The company’s chief research officer, Mark Chen, told reporters that the computational effort ran emphatically in the millions of dollars
and required deploying 10,000 autonomous AI agents working simultaneously.
Independent Researchers Allege OpenAI Built on Their Work
The announcement immediately collided with allegations from New York University professor Tristan Buckmaster and Anthropic researcher Levent Alpöge. As reported by USA Today, the two mathematicians spent nearly a year working independently on fluid equations, building upon prior research established by mathematicians Diego Córdoba and Luis Martínez-Zoroa.
Buckmaster published a statement alleging that after OpenAI learned of their progress through informal channels, the company raced to solve the Navier-Stokes problem using a similar route. MIT Technology Review noted that Buckmaster posted a proof on Mastodon showing a simplified version of the equations breaking down just before OpenAI presented its solution for the full equations.
“The significance of this with respect to the way we train students, assign credit, referee, and decide what is worth one human life’s attention cannot be understated.”
Tristan Buckmaster, New York University professor, via USA Today
The Dispute Over Codex Transcripts and Authorship
Central to the controversy are questions regarding training data and user privacy. Buckmaster stated that after hearing rumors about OpenAI’s work, he contacted an OpenAI representative and subsequently spoke with Sebastian Bubeck. During that call, Buckmaster asked whether OpenAI’s model had access to or training on the sessions he and Alpöge conducted inside Codex, their coding agent workspace.
Buckmaster received assurances that the model did not look up user data, though he received no direct answer regarding model training on those transcripts. Furthermore, Buckmaster claimed Bubeck proposed publishing their findings together on the condition that Alpöge be excluded from authorship due to his employment at Anthropic, OpenAI’s rival.
OpenAI Denies Access While Acknowledging Training Realities
OpenAI formally denied that the researchers’ work was viewed to achieve the breakthrough. In statements covered by USA Today, the company insisted its researchers did not see any of their work through any means until they released it publicly — in particular, no specific user data was accessed in order to solve this problem.
Bubeck explained that OpenAI initiated its Millennium problems effort in response to viral Twitter rumors that Anthropic had resolved two such problems.
However, OpenAI subsequently admitted it could not entirely rule out the possibility that de-identified data derived from product usage helped improve its models overall. Mark Chen reiterated during a press briefing that agents did not access user transcripts, though MIT Technology Review pointed out that auditing autonomous agents remains an evolving challenge for AI labs.
The Long Road to Verification and Prize Rules
Despite the rapid computational output, claiming the official $1 million prize from the Clay Mathematics Institute is far from immediate. Professor Martin Bridson, President of the Clay Mathematics Institute, told AFP that prize rules mandate formal publication in a peer-reviewed journal followed by a mandatory two-year waiting period for community scrutiny before the institute even convenes a committee to evaluate the work.
The process of evaluation is deliberately unhurried, and we shall ensure that it is absolutely rigorous,
Professor Martin Bridson stated, while OpenAI confirmed it will not claim the cash reward itself if the proof is ultimately validated.
