NYU Mathematician Accuses OpenAI of Racing to Publish Millennium Problem Proof
The Millennium Prize Problems, established by the Clay Mathematics Institute, rank among the most consequential open questions in mathematics, with a $1 million award attached to each. An NYU mathematics professor and an Anthropic researcher used AI systems including Claude in their work, making the dispute a test of how academic credit, confidential findings and model-assisted research should be handled as technology companies compete for scientific breakthroughs.
The researchers accused OpenAI of drawing on their work before it was publicly released and rushing out a complete proof. OpenAI disputed that account, saying an unreleased next-generation model produced the proof independently using $22.5 million of computing resources. The conflicting claims have prompted scrutiny of research timelines, source attribution and access to unpublished material, while the mathematical community assesses both the proof itself and whether accepted standards of scholarly conduct were followed.
All Coverage
2 original reportsThe Backstory
The history behind this eventOpenAI Rejects Plagiarism Claims Over Math Breakthrough
The Navier-Stokes regularity problem asks whether equations governing viscous fluids can develop singularities in three dimensions. It is one of the Clay Mathematics Institute’s seven Millennium Prize Problems, carrying a $1 million award. OpenAI says an internal model produced a proof that blowups can occur, a potentially historic result that would also test how credit and provenance should be assigned when artificial intelligence participates in original research.
OpenAI announced the Lean-verified proof on Sept. 8, 2026. New York University mathematician Tristan Buckmaster and Levent Alpöge, a mathematician employed by Anthropic, alleged that the lab pursued a similar “forcing” approach after learning of their related work and completed its proof over the weekend. Chief Executive Sam Altman called the plagiarism allegations unfounded, while OpenAI said no specific user data or unpublished work from the pair was accessed. Independent scrutiny of the competing proofs and their provenance is continuing.
Subscribe to Mark Radar Weekly
Every Friday, the week's strongest signals in your inbox. Unsubscribe anytime.
If you search news on Google, you can set Mark Radar as a preferred source—our coverage will show up more often in your results. Set as preferred source on Google →