theverge.com web signal

OpenAI's Navier-Stokes proof triggers credit fight with NYU

TL;DR

  • OpenAI posted its Navier-Stokes proof roughly twelve hours after NYU's Tristan Buckmaster released his own claim on the same Millennium Prize problem.
  • Buckmaster says OpenAI's Sébastien Bubeck offered credit if he dropped his Anthropic collaborator Levent Alpöge, calling it a bribe.
  • OpenAI's 88-hour effort used around 10,000 agents and about $22.5 million in compute, and cannot rule out training benefit from de-identified data.

OpenAI's blog post announcing it had solved the Navier-Stokes Millennium Prize Problem went live roughly twelve hours after Tristan Buckmaster, a mathematician at New York University's Courant Institute, released his own statement claiming the same result. According to The Verge, OpenAI ran the effort with about 10,000 agents over 88 hours, at a compute bill the company put near $22.5 million.

The dispute is not about the math. It is about how OpenAI got there.

Buckmaster and his collaborator Levent Alpöge, a mathematician at Anthropic, say OpenAI approached them after learning of their progress and pushed a deal that would have cut Alpöge out of the credit. "All I had to do was throw Levent under the bus," Buckmaster said. He alleges that on a September 6 call, Sébastien Bubeck, the OpenAI researcher leading the project, told him "Why would you ruin your career?" and later, "If you don't want me to be nice, then I don't have to be nice."

Bubeck wrote on X that "a series of false and inflammatory allegations against me are currently circulating on social channels." OpenAI's own post conceded a narrower point about training exposure: the company said it "cannot rule out that de-identified data...helped improve our models." That hedge matters because Buckmaster had been using OpenAI's Codex tool during his work.

Terence Tao framed the discomfort in ecological terms. "The indiscriminate strip-mining of open problems for solutions may destroy the ecosystem from which the next generation of mathematical techniques, problems, and practitioners would have developed," he wrote. Nearly 3,900 mathematicians have signed the Leiden Declaration opposing AI hype and corporate intrusion into mathematics research; two researchers we track shared the Verge piece within hours of publication.

The $1 million Clay Mathematics Institute bounty is the sideshow. The bigger question is procedural: whether a lab that runs its own coding tool, its own agents, and its own compute budget can credibly claim independent priority on a problem that a handful of humans were visibly closing in on.

Shared on Bluesky by 2 AI experts