OpenAI's Math Breakthrough Marred by Credit Dispute
OpenAI announced Tuesday that an internal AI model proved the three-dimensional Navier-Stokes equations can break down under certain conditions. The claim arrived alongside allegations from an outside mathematician that OpenAI copied his research direction after learning of his unpublished work.
- 10,000 AI agents ran concurrently on the problem, reaching a result in about 88 hours.
- Computing costs ran 'emphatically in the millions of dollars,' roughly 1,000 times previous math efforts.
- OpenAI says it will not claim the $1 million prize if the result is confirmed.
- NYU mathematician Tristan Buckmaster and Anthropic researcher Levent Alpöge released related findings the same day.
Why it matters: The controversy centers on a core trust problem for AI-assisted science: researchers using commercial AI tools to develop unpublished work may inadvertently train the same systems to compete against them.
- OpenAI's terms of service allow training on user-submitted data unless users opt out through its privacy portal.
- Buckmaster said his team used OpenAI's Codex extensively throughout their year-long project but did not specify whether they opted out of model training.
How 12 sources split on this story
Where they split: The central dispute is whether OpenAI independently arrived at Buckmaster and Alpöge's specific research approach or learned of it — directly or through training data — before racing to a solution with vastly greater computing resources.
Center coverage, 7 sources: The center treats both the mathematical achievement and the credit dispute as equally significant, focusing on what the episode reveals about the pace of AI progress and the structural risks of AI-assisted research.
Axios8hOpenAI's historic math solution overshadowed by credit controversy
Fortune5hOpenAI says it cracked Navier-Stokes, one of math's grand challenges. | FortuneCNCTV News4hOpenAI says AI solved one of math’s hardest problems in days
Engadget2hWhat's Going On With OpenAI And The Navier-Stokes Controversy?
Financial Times36mOpenAI faces competing claims around maths breakthrough
Semafor2hOpenAI agents find proof to $1 million Millennium Prize ProblemYNYahoo News4hOpenAI says it cracked 90-year-old maths problem in 88 hoursLeft coverage, 5 sources: The left frames the story as a cautionary tale about corporate AI power — a well-funded lab potentially exploiting an individual researcher's unpublished work and then pressuring him to stay quiet.
TechCrunch7hOpenAI fought dirty on career-making math problem, says NYU mathematician | TechCrunch
Business Insider5hOpenAI's big math breakthrough claim sparks drama
The New York Times7hWhat Is Navier-Stokes, the ‘Millennium’ Math Problem OpenAI Claims to Have Solved?
The Guardian3hOpenAI claims to have solved maths problem that stumped humans for decades
The Verge4hDrama swirls around OpenAI’s legendary mathematical milestoneWhat’s next: Clay Mathematics Institute President Martin Bridson said evaluation will be 'deliberately unhurried' and 'absolutely rigorous.'
- OpenAI's Sebastian Bubeck said a fuller statement addressing the allegations would follow.
- Did Buckmaster and Alpöge's Codex interactions — through training or direct access — influence the approach OpenAI's model ultimately used?
- Will peer reviewers confirm OpenAI's proof is complete, given that Buckmaster and Alpöge stopped short of a full solution using the same approach?
- How should scientific credit be apportioned when AI models trained on one researcher's work help a rival reach a result first?