Scientific American on OpenAI’s Navier-Stokes claim amid controversy
Scientific American ● Covered by 18 sources
OpenAI says its AI proved a Millennium Prize math problem about fluids. Mathematicians are already arguing over whether it got there by its own route.
Based on reporting by Scientific American — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI has thrown a very loud claim into one of math’s most famous open problems. The company says its internal model has proved a flaw in the Navier-Stokes equations, the formulas mathematicians use to describe how fluids move. If that holds up, it would mark only the second time anyone has solved one of the seven Millennium Prize Problems, each carrying a $1-million reward.
The specific issue is whether the equations can ever “blow up.” OpenAI says its proof shows that, in rare cases, the math predicts a fluid’s speed becoming infinite at some points — a result that cannot exist in the real world. The company also says the proof has been checked in Lean, the programming language used to certify formal mathematics, which is about as close as this field gets to a machine-backed stamp of correctness.
But the announcement landed in the middle of a dispute over who got there first, and how. Mathematician Tristan Buckmaster said OpenAI may have moved on the problem after learning that he and Levent Alpöge, a mathematician at Anthropic, had been using a specific method on a related question. Buckmaster posted the night before OpenAI’s announcement that the pair had blown up the Euler equations, a step many researchers see as closely tied to Navier-Stokes.
OpenAI’s Sébastien Bubeck pushed back hard. He said the company’s model had independently solved the Euler problem by different means, while the Navier-Stokes proof itself followed a similar broad method to the one Buckmaster and Alpöge were using. He also said the work on the full proof was done over the weekend, after the time Buckmaster says OpenAI may have heard about the earlier result. OpenAI flatly denied using the other team’s prompt or proofs.
For now, the math community is left with two things at once: a potentially historic result, and a fight over provenance that may be impossible to untangle quickly. Luis Silvestre of the University of Chicago put it plainly: the last two days have been crazy, and everyone is already talking about what it means.
My take — AI-written commentary, not fact-checked reporting
This is exactly why AI math claims need more than glossy victory laps. A proof that matters is one the field can inspect, not one wrapped in startup adrenaline and a weekend timeline. If the model really found a new route, great; if not, this will be remembered as another case of tech treating provenance like a minor formatting issue.
Read more about this at: Scientific American
