OpenAI claims to have solved a major math problem, but faces backlash from NYU's Tristan Buckmaster over alleged unethical practices.
Key Points
- OpenAI announced a solution to the Navier-Stokes equations, claiming it was verified by AI agents over 88 hours.
- NYU mathematician Tristan Buckmaster accused OpenAI of unethical practices related to his unpublished proof.
- OpenAI and its representatives denied the allegations, asserting integrity in their actions and collaboration offers.
OpenAI recently claimed it has solved a critical mathematical question regarding the Navier-Stokes equations, a significant problem in fluid dynamics. However, the announcement has sparked controversy following accusations from NYU mathematician Tristan Buckmaster, who alleges that OpenAI engaged in unethical behavior regarding credit for the proof.
The Navier-Stokes equations are among the seven Millennium Prize Problems, each offering a $1 million reward for a valid solution. OpenAI stated that its solution was achieved through collaboration among around 10,000 AI agents over a span of 88 hours, with validation from Lean, a software used for checking mathematical proofs.
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.
The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.
However, Buckmaster claims that he and mathematician Levent Alpöge at Anthropic had already been working on a similar proof for nearly a year, achieving their solution on August 22. Just before OpenAI’s announcement, Buckmaster revealed details about their work and alleged that an OpenAI mathematician was informed about their progress.
In a phone conversation shortly after, Buckmaster alleges that OpenAI's Sébastien Bubeck mentioned they had also developed a proof and presented two options: publish alongside them or publish alone, excluding Alpöge due to his affiliation with a competitor. Buckmaster chose to make his findings public, which led to Bubeck questioning his decision.
After Buckmaster published his findings, which recognized Alpöge's contributions, OpenAI did not acknowledge their work, only stating it could not rule out that their models were influenced by de-identified data from users of their products.
OpenAI and Bubeck Respond
In response to the allegations, Bubeck and OpenAI have both denied any wrongdoing. Bubeck shared messages indicating an intention to coordinate with Alpöge and insisted he acted with integrity throughout the process. OpenAI CEO Sam Altman supported Bubeck, stating that the company originally believed Buckmaster's team had also solved the problem and aimed for a collaborative release.
Initially we believed the other team had also solved the problem. We wanted to collaborate and do a joint release.
OpenAI later clarified that it had not seen Buckmaster and Alpöge's work prior to their publication, although it did not completely dismiss the possibility of using data from their products. Notably, Fields Medalist Terence Tao commended the work of Buckmaster and Alpöge as a significant achievement.
This incident follows another notable development in AI and mathematics, as Anthropic recently claimed that its AI, Claude, produced a verified proof of Fermat's Last Theorem. As of now, Buckmaster and Alpöge's findings remain unpublished pending final verification with Lean, while OpenAI's 100-page proof has not yet undergone external review.
