OpenAI announced that an unreleased model solved the Navier-Stokes equation, a Millennium Prize problem unsolved for nearly ninety years, after 88 hours of computation by about ten thousand AI agents, sparking tension with academics who allege the firm pressured them to share credit.
Short answer: OpenAI announced that an unreleased model solved the Navier-Stokes equation, a Millennium Prize problem unsolved for nearly ninety years, after 88 hours of computation by about ten thousand AI agents, sparking tension with academics who allege the firm pressured them to share credit.
OpenAI announced on Tuesday that one of its unreleased models had cracked the Navier-Stokes equation, a Millennium Prize problem that’s stumped mathematicians for almost ninety years. The breakthrough came after just 88 hours of number-crunching, powered by a swarm of roughly ten thousand AI agents steered by an internal orchestration system. OpenAI framed the result as a milestone that shows how fast AI is reshaping mathematical research.
The timing, however, has ruffled a few feathers in academia. Just a day before OpenAI’s statement, New York University professor Tristan Buckmaster - together with Levent Alpöge from Anthropic - posted work on a related problem. Buckmaster said he reached out to OpenAI after learning they’d caught wind of their progress, asking when the effort kicked off and what data had trained the model. According to him, the exchange quickly turned sour. An OpenAI representative reportedly asked why he’d risk his career by going public, and when he pressed for clarification, the reply was that if he didn’t want niceness, then none would be offered. OpenAI then nudged Buckmaster to publish the findings while crediting the company’s model and dropping Alpöge as a co-author.
Buckmaster also wondered whether OpenAI had peeked into his private sessions on Codex, the tool he used while tackling the problem. He said the company’s replies grew evasive and at times downright hostile. In later statements, OpenAI denied any direct use of user data, insisting its researchers and agents hadn’t seen the pair’s work before it went public. The firm did concede it couldn’t entirely rule out an indirect influence, noting that de-identified information harvested from users’ interactions with its products might have fed model improvements, although it stressed that the two proofs differ substantially.
OpenAI described the effort as hurried and expensive, claiming it cost millions of dollars. Yet the firm said it has no intention of chasing the $1 million bounty administered by the Clay Mathematics Institute - which, by the way, hasn’t awarded the prize yet. Instead, OpenAI said its goal was simply to highlight the progress of its AI systems. The company added that it hadn’t publicly worked on Navier-Stokes before September, nor had it discussed any prior attempts.
The rush to tackle the problem reportedly started after OpenAI caught wind of rumors on Twitter that other researchers were making headway on Millennium Prize questions. Sébastien Bubeck, an OpenAI researcher, told a press briefing covered by Science that the team wondered why not try to solve one themselves given the strength of their model. Only later did they realize the rumors were about Alpöge and Buckmaster.
Even without the most explosive accusations, aspects of OpenAI’s conduct that the company itself has acknowledged have shocked many mathematicians. The apparent attempt to beat competitors to a result runs counter to the discipline’s usual pace, where deep, specialized expertise often keeps outsiders from swooping in. Openness and collaboration are longstanding norms, and the episode has raised concerns that a more aggressive, prize-driven approach could chill the collaborative spirit that has long defined mathematical inquiry. Researchers are now watching closely to see how this will shape future interactions between AI labs and the academic community.
Yes, OpenAI announced that one of its unreleased models had cracked the Navier-Stokes equation, a Millennium Prize problem that has challenged mathematicians for nearly ninety years, after just 88 hours of computation using roughly ten thousand AI agents.
No, OpenAI said it has no intention of chasing the $1 million bounty administered by the Clay Mathematics Institute, noting that the prize has not yet been awarded and its goal was merely to showcase the progress of its AI systems.
OpenAI denied any direct use of user data, stating its researchers and agents had not seen Buckmaster and Alpöge’s work before it went public, but conceded it could not entirely rule out an indirect influence from de-identified information harvested from users’ interactions with its products.
Buckmaster said he reached out to OpenAI after learning they had caught wind of their progress, asking when the effort started and what data trained the model, but the reply turned sour, with an OpenAI representative asking why he would risk his career by going public and stating that if he did not want niceness, none would be offered.
Source: The Verge
Microsoft announced on September 9, 2026 that it has agreed to a set of AI safety and privacy principles for use in US schools, pledging not to train its models on student or educator data, limiting data collection, providing plain-language explanations to families, banning AI companions that interact directly with learners, and requiring human review for high-risk decisions, with districts able to opt-in starting in November.
On September 9, 2026, OpenAI announced that its unreleased AI model solved the Navier-Stokes Millennium Prize problem in 88 hours using a swarm of about ten thousand internal agents, a claim that sparked controversy after NYU professor Tristan Buckmaster alleged the company pressured him to drop co-author Levent Alpöge and questioned its use of his data.
Suno unveiled its sixth-generation AI music model, V6, on September 9 2026, marking the first time it collaborated directly with major record labels. The model draws on licensed tracks from Warner Music Group, BMG and Believe, plus user-contributed audio.
NO COMMENTS YET
Comments are open. Have a thought or a question? Share it below.