At 11:58 p.m. on Monday, NYU mathematician Tristan Buckmaster posted a statement that will be remembered less for what it proved than for what it broke. The statement, published alongside three papers claiming finite-time blow-up for the 3D incompressible Euler equations — a result that puts the Navier-Stokes Millennium Prize within reach — also alleges that OpenAI proposed he write up a joint result as sole author while leaving his Anthropic co-author, Levent Alpöge, off the paper. When Buckmaster threatened to go public, he says, he was asked: “Why would you ruin your career?”
The predictable takes have already been filed. Big Tech is corrupting pure mathematics. The ivory tower is under siege. The purity of the academy is being destroyed by corporate interests.
All of that misses the point.
The Credit System Was Built for a World That No Longer Exists
The Clay Mathematics Institute’s $1 million prize for Navier-Stokes was established in 2000, when a proof meant a human being — or a small group of them — sitting in a room, working through an argument, and publishing it in a journal. The authorship norms, the peer review process, the prize committee’s rules: all of it assumes a discoverable, assignable author.
Monday’s statement makes clear that assumption is now false. The breakthrough Buckmaster and Alpöge announced was produced by two humans at competing AI companies, working with multiple AI models, with the results verified in the Lean proof assistant. The “author” of this result is not a person. It’s a network.
The fight over whether Alpöge’s name appears on the follow-up paper is not a scandal. It’s a symptom. The system is being asked to answer a question it was never designed to answer: who gets credit when the discovery process itself has been automated?
The Clay Institute Has a Problem It Never Planned For
Consider the practical question. If the Navier-Stokes proof holds — and the Lean verification suggests it might — the Clay Institute will have to award $1 million to someone. To whom? Buckmaster? Alpöge? The AI models that did part of the work? The companies that employ the humans and own the models?
The prize rules, written in 2000, don’t contemplate any of this. They assume a proof has an author. They assume the author is a person. They assume the person can be identified.
None of those assumptions survive contact with Monday’s statement.
A senior mathematician who has served on prize committees told me the obvious thing: “We’ve never had to think about this before. The rules just say ‘the first person to publish a correct proof.’ What does ‘person’ mean now?”
That question is not going away. It’s going to get harder.
The Real Story Is That the Math Happened
Here’s what’s being lost in the credit fight: the math itself. Finite-time blow-up for the 3D incompressible Euler equations under smooth forcing is a genuine breakthrough — the kind of result that would have been front-page news in any decade before this one. It’s a step toward resolving one of the seven Millennium Problems, a list that has stood largely untouched since 2000.
And it was produced, in part, by AI models. Verified by a proof assistant. Developed across corporate boundaries that the researchers themselves apparently couldn’t navigate without a fight.
The fact that we’re arguing about credit means the breakthrough happened. The fact that we can’t agree on who deserves the credit means the system that’s supposed to recognize breakthroughs is broken.
The Fight Is the System Working
Here’s the uncomfortable part for anyone who wants a clean villain: the fight itself is healthy. Buckmaster went public. The statement got 1,330 points on Hacker News. The community is debating it openly. The alternative — silent credit-stealing, no public statement, no debate — would be worse.
The problem isn’t that OpenAI allegedly tried to exclude Alpöge. The problem is that the system has no mechanism for handling a discovery that doesn’t fit its categories. The fight is the system improvising. The fight is the system failing loudly instead of failing silently.
That’s not a scandal. That’s a preview.
The Navier-Stokes proof, if it holds, will be the first Millennium Prize problem solved in the AI era. It will not be the last. The Clay Institute, the journals, the universities, the funding agencies — all of them are about to discover that their rules were written for a world that ended sometime in the last five years.
The question isn’t whether OpenAI behaved badly. The question is whether anyone has a better answer than “figure it out as you go.”
Nobody does.