OpenAI announced in September 2026 that it has found a solution to the Navier-Stokes problem, a major mathematics challenge that has remained unsolved for roughly 90 years. The discovery was made using an internal AI model described as more powerful than the recently released GPT-6 Astra, running alongside 10,000 concurrent agents.
The Navier-Stokes problem, which relates to the flow of liquid and gas, is one of seven Millennium Prize Problems — each carrying a $1 million reward for a verified solution. OpenAI says it does not plan to claim the prize money. The company began training its internal model on August 28th, saying it has “exhibited unprecedented performance in our benchmarks, including mathematics.”
The announcement has been met with controversy. One day before OpenAI’s blog post, New York University mathematics professor Tristan Buckmaster published findings on a related problem, produced in partnership with Levent Alpöge, a researcher at Anthropic. Buckmaster claims he contacted OpenAI after learning the company had become aware of his team’s progress, and alleges that OpenAI produced a proof using a route he and Alpöge had been developing — work they had been inputting into OpenAI’s Codex tool throughout the project.
Buckmaster raised direct concerns about whether OpenAI had accessed their Codex session data. “I asked whether the model had been trained on, or had access to, our sessions in Codex, into which we had been putting all our drafts for the whole of this project,” he wrote. “I was told the model did not look up user data. I asked again, about training, and I did not get an answer.”
OpenAI responded by stating that “no specific user data was accessed in order to solve this problem,” while also acknowledging that it “cannot rule out that de-identified data derived from their usage of our products helped improve our models.” OpenAI technical staff member Sebastien Bubeck added that the company “did not see any of their work until they released it publicly,” and that the two proofs “differ significantly.”
Buckmaster pushed back on Mastodon, claiming OpenAI is “openly admitting they used training data from a period after we found our result.” The dispute raises questions about data privacy for users of AI coding and research tools, and about how credit for major scientific discoveries may be assigned when AI systems are involved.
Source: The Verge