OpenAI's AI Solved a Famous Math Problem. Then the Accusations Started

On Tuesday, OpenAI announced that one of its unreleased AI models solved the Navier-Stokes problem — a set of equations describing how water, air, and blood actually flow. It's one of the famous Millennium Prize problems, with a $1 million bounty for whoever cracks it, and it has stumped human mathematicians for roughly 90 years. OpenAI says its model did it in 88 hours by directing about 10,000 AI agents — think of them as thousands of tireless research assistants each chewing on one slice of the puzzle at once.

The timing, though, is what upset people. Just one day earlier, New York University professor Tristan Buckmaster published related findings with Levent Alpöge, a researcher at Anthropic — OpenAI's biggest rival. Buckmaster asked OpenAI when it started work on the problem and what data its model learned from. He says the conversation turned nasty. An OpenAI researcher allegedly asked 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." He also says OpenAI pushed him to publish the work while crediting OpenAI's model and dropping his coauthor.

OpenAI flatly denies any wrongdoing. In its blog post, the company says its researchers and its AI never saw the rival team's work before it was published, and that no specific user data was accessed. But it added a caveat: while "unlikely," it cannot fully rule out that anonymous, de-identified data from people using its products helped improve its models. That admission is exactly what worries academics, because it's nearly impossible to trace where an AI's ideas originally came from.

So why should you care? Because this is a preview of how AI is about to change science, medicine, and engineering — sometimes by racing ahead of the humans who've spent years on a problem. Beyond bragging rights and prize money, the fight raises real questions about who gets credit for discoveries, whether your use of AI tools quietly feeds someone else's breakthrough, and whether researchers will start guarding their work instead of sharing it.