OpenAI Claims AI Solved Math Problem, Mathematician Accuses Research Plagiarism

OpenAI has claimed that an AI system solved the Navier-Stokes problem, one of mathematics' seven Millennium Prize Problems considered among the most difficult unsolved challenges.
The company stated that 10,000 AI agents collaborated for 88 hours to complete the proof. NYU professor Tristan Buckmaster disputes this claim, stating he and his team deserve credit for similar results.
The Navier-Stokes equations describe fluid motion in air or water. OpenAI's proof suggests these equations could reach a 'finite-time blowup' where fluid velocity becomes unbounded.
OpenAI did not disclose the model used, but stated it outperforms the publicly available GPT-6 Astra system. GPT-6 Astra later verified the logic of the proof over 17 hours.
The Clay Mathematics Institute offers a $1 million prize for solving one of the seven Millennium Prize Problems. OpenAI has declined to apply for this reward.
NYU professor Tristan Buckmaster and mathematician Levent Alpöge from Anthropic claim they were close to similar results, with their research reportedly incorporated into OpenAI's Codex coding AI model. Buckmaster argues this exposure may have influenced OpenAI's team.
OpenAI researcher Sebastien Bubeck stated the company did not access or use unpublished research. However, OpenAI acknowledged that product usage by researchers might have affected its models' training data.
The dispute comes as OpenAI advances a $1 trillion IPO plan, while Anthropic pursues an independent IPO aiming for higher valuation.
Korean Source
This article is an English localization of a Korean-language crypto news report. Original headline: 오픈AI, 수학 난제 해결 주장…수학자 ‘AI가 연구 도용’