OpenAI, NYU mathematician clash over AI-assisted Navier-Stokes math breakthrough

NYU mathematics professor Tristan Buckmaster announced three proofs with a preliminary finding tied to the Navier-Stokes existence and smoothness problem. The work was done with Anthropic mathematician Levent Alpöge and used both OpenAI’s Codex and Anthropic’s Claude models. The scientific claim quickly became entangled with a dispute over whether OpenAI’s parallel work drew on nonpublic information about their progress.

The Navier-Stokes existence and smoothness problem is one of the seven Millennium Prize problems. Each carries a $1 million bounty from the Clay Mathematics Institute for the first person or group to provide a solution. The equations are widely used in fluid mechanics, but remain difficult in theoretical terms, making any credible progress highly significant for mathematical physics.

After Buckmaster’s statement, OpenAI published what it described as a full proof of the Navier-Stokes problem, saying it was discovered by an unreleased next-generation model. TechCrunch reports that OpenAI said the effort used 300 billion output tokens, equal to $22.5 million worth of compute if charged at current Astra rates. OpenAI’s post said its latest effort began on September 1 after rumors that two Millennium Prize problems had been solved.

Buckmaster alleged that information about his and Alpöge’s progress reached OpenAI before it became public, and that OpenAI’s answers about timing and human involvement became evasive. He also raised concerns that his use of Codex could have exposed elements of the work to OpenAI’s systems, while OpenAI said no specific user data was accessed to solve the problem and downplayed the likelihood of regurgitation. The dispute is likely to intensify debate over how AI labs should handle credit, private research signals, and model-training data in high-stakes academic work.