← Back to articles

OpenAI Says Its AI Just Solved a 90-Year-Old Math Problem — One Mathematician Calls It Theft

2026-09-09 | original

read original get Princeton University Press "The Millennium Problems" by Keith Devlin → more articles

Why This Matters

A dispute over OpenAI's claim that its AI resolved a 90-year-old Millennium Prize problem highlights how quickly AI is colliding with academic norms of credit and attribution. Mathematician Tristan Buckmaster alleges OpenAI built on unpublished work he developed with Anthropic's Levent Alp\u00f6ge, and that credit was offered only if his collaborator at a rival lab was excluded. The episode matters because AI research claims are becoming marketing, and the rules for provenance and authorship haven't caught up.

Key Takeaways

OpenAI says its AI solved a long-standing fluid-equations problem in days using 10,000 agents; Buckmaster says it leaned on his unpublished approach.
Buckmaster alleges OpenAI conditioned credit on excluding co-author Levent Alp\u00f6ge because he works at Anthropic, plus a pointed remark about his career.
The fight underscores the lack of clear attribution norms for AI-assisted results and the competitive pressures shaping how labs publicize breakthroughs.

Worth a Look

Princeton University Press "The Millennium Problems" by Keith Devlin — If this fluid-equations drama has you curious about the Clay Institute's famous seven, Keith Devlin's book walks general readers through each problem and why they've resisted proof for decades. It's a great way to understand what the Navier-Stokes question actually asks before arguing about who solved it.

See Princeton University Press "The Millennium Problems" by Keith Devlin on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.

Listen to this post

When OpenAI cracked one of math’s six unsolved “million-dollar” problems, the academic world was stunned. The 90-year-old question, part of the Clay Mathematics Institute’s famous Millennium Prize Problems, asks whether the equations that govern fluid motion can break down in extreme situations. OpenAI says its AI resolved it in a matter of days. But mathematician Tristan Buckmaster says OpenAI copied off of his work, according to Scientific American.

In a statement, Buckmaster said he and fellow mathematician Levent Alpöge, who works at Anthropic, had already made major progress on the problem. But before they could publish, he says word of their approach reached OpenAI, and the company used it to finish the job over a single weekend with the help of 10,000 agents.

Buckmaster says things got weird after that. He claims OpenAI offered to give him full credit, but only if Alpöge, who works for a competitor, was left out entirely. When he pushed back, he says an OpenAI employee told him, “Why would you ruin your career?”

Explore topics: openai tristan buckmaster clay mathematics institute millennium prize problems levent alpöge