Who is Tristan Buckmaster? NYU mathematician at centre of OpenAI Navier Stokes controversy over private Codex logs and research credit

Tristan Buckmaster was already a big name in mathematics long before his name became attached to one of the biggest AI controversies of 2026. The NYU mathematician is a leading expert on fluid equations and has spent years studying some of the hardest questions around turbulence, singularities and the behaviour of fluids.


Now, his work has become part of a very different fight involving OpenAI, private Codex sessions, research credit and its claimed solution to the famous Navier Stokes problem.

Buckmaster is a professor of mathematics at New York University’s Courant Institute of Mathematical Sciences. He completed his PhD at the University of Leipzig and the Max Planck Institute for Mathematics in the Sciences in 2014, before spending several years at NYU and Princeton and later becoming a professor at the University of Maryland. He returned to NYU as a professor and continues to work on problems involving fluid dynamics and singularity formation.

His reputation is not built overnight. In 2019, Buckmaster shared the prestigious Clay Research Award with Philip Isett and Vlad Vicol for major contributions to the analysis of partial differential equations, particularly the Euler and Navier Stokes equations. His work with Vicol showed that weak Navier Stokes solutions could behave in surprisingly wild and non unique ways.


The answer is Navier Stokes, one of the seven Millennium Prize Problems. In simple terms, mathematicians have spent decades trying to determine whether smooth three dimensional fluid equations can develop a singularity, where their behaviour effectively breaks down.

OpenAI announced on September 8 that an internal AI system had produced a proof showing finite time singularity formation. The company said around 10,000 coordinating AI agents worked on the problem for about 88 hours, with the resulting argument also formalised in Lean.

But there was a problem. Buckmaster and Anthropic mathematician Levent Alpöge had been working on a closely related route, using AI tools including OpenAI’s Codex. Buckmaster said information about their unpublished work reached OpenAI shortly before the company began its own intensive effort.

This is where the story gets messy. Tristan Buckmaster’s allegation is more nuanced than saying OpenAI simply “stole” his research. He says he and Levent Alpöge had spent nearly a year working on a specific route toward the Navier Stokes problem, with drafts, calculations and research notes stored in private Codex sessions. According to Buckmaster, OpenAI began its intensive effort after learning about their progress and eventually produced a proof using a closely related technical direction.


But Buckmaster has not seen OpenAI’s proof and has repeatedly said he does not know whether his private data was accessed or used. His concern is based on the timing, the similarity of the approach and what he says he was told during conversations with OpenAI researcher Sébastien Bubeck.

The dispute also became personal over authorship. Buckmaster says Bubeck wanted Alpöge left out because he works for Anthropic, which competes with OpenAI. He also alleges that when he threatened to make the dispute public, he was asked, “Why would you ruin your career?” Bubeck has disputed Buckmaster’s account and called the allegations “false and inflammatory.”

OpenAI has pushed back strongly against the account. The company says its researchers did not access Buckmaster and Alpöge’s private work before it was publicly released and maintains that its proof was developed independently.

OpenAI researcher Sébastien Bubeck has also disputed Buckmaster’s version of events, saying he never asked for Alpöge to be removed from authorship of the researchers’ own work. He has argued that the authorship discussion concerned whether an Anthropic employee could participate in an OpenAI paper.


They have nevertheless acknowledged a wrinkle that has kept the Codex question alive. The company says it cannot completely rule out the possibility that de-identified data from product usage could have contributed indirectly to model improvement. That is very different from saying the model directly accessed Buckmaster’s private sessions.

The biggest question now is evidence. OpenAI says its researchers and agents did not access Buckmaster and Alpöge’s unpublished work before it became public, although the company has acknowledged that it cannot completely rule out deidentified product data indirectly contributing to model improvements.

Meanwhile, mathematicians will have to examine OpenAI’s proof and determine whether it is correct and actually meets the requirements of the Clay Mathematics Institute’s Navier Stokes problem. The overlap between the two projects, the timeline and the Codex data question will likely face scrutiny as well.

For now, there is no public proof that OpenAI accessed or used Buckmaster’s private Codex logs. What exists is a serious dispute over timing, research credit, data practices and the circumstances surrounding one of the most extraordinary AI mathematics claims yet.