OpenAI Accused of 'Dirty' Tactics in Race to Solve $1M Math Problem

NYU mathematician Tristan Buckmaster has accused OpenAI of using unfair tactics to scoop his team's work on the $1 million Navier-Stokes Millennium Prize problem. Buckmaster claims OpenAI leveraged massive compute resources and potentially his own Codex usage data to rush a full proof after learning of his team's progress. OpenAI denies the allegations, asserting their model arrived at the solution independently despite being inspired by rumors of the academic team's success.

Key Points

• NYU's Tristan Buckmaster and collaborator Levent Alpöge developed a preliminary proof for the Navier-Stokes problem using AI assistance.
• OpenAI released a full solution shortly after, claiming it was achieved by a next-gen model using $22.5 million worth of compute.
• Buckmaster alleges OpenAI used 'dirty' tactics, including potentially training on his Codex interaction data and rushing to finish once they heard of his progress.
• OpenAI denies direct plagiarism or data misuse but acknowledges starting their effort after hearing rumors of a solution.
• The conflict highlights the tension between academic research and the massive compute power and data access of corporate AI labs.

Sentiment

The overall sentiment is negative toward OpenAI and broadly sympathetic to the article's criticism, though not uniformly convinced that outright plagiarism occurred. Commenters tend to agree that the company's behavior looks opportunistic and poorly aligned with academic norms, while a smaller group argues that competitive AI research on a major open problem is fair game. The strongest consensus is that the episode exposes unresolved problems around credit, transparency, and trust when commercial AI labs enter frontier scientific research.

In Agreement

• OpenAI appears to have used non-public signals about another team's progress to rush its own effort, which many commenters see as ethically dubious even if the proof is technically independent.
• The company had overwhelming compute and institutional advantages, creating an unfair race against researchers who had spent far longer developing the mathematical direction.
• The alleged pressure on Buckmaster to avoid going public is viewed as especially troubling and consistent with broader distrust of OpenAI's ethics and transparency.
• Academic credit should account for intellectual direction, intermediate work, and provenance, not only the final text of a proof produced by a model.
• The episode may push researchers to hide work from AI labs and commercial AI tools, damaging open scientific collaboration.

Opposed

• OpenAI's stated explanation could be valid: it believed a competitor-linked researcher had already solved the problem and was testing whether its own models could match that result.
• Solving a famous mathematics problem is legitimate research, and competition between labs can accelerate progress rather than merely violate academic etiquette.
• If OpenAI's proof and the Buckmaster-Alpoge work are materially different, accusations of theft or plagiarism may be overstated.
• Some commenters argue that no company or individual has exclusive standing to pursue a major open problem, regardless of who was close first.
• The result still demonstrates impressive potential for AI-assisted mathematics, even if the surrounding judgment and attribution were mishandled.