OpenAI announced that its internal agents have solved the Navier–Stokes existence and smoothness problem, one of the seven Millennium Prize Problems in mathematics. However, the announcement has been overshadowed by accusations that the company failed to credit the prior work of NYU mathematician Tristan Buckmaster and Anthropic employee Levent Alpöge, who were also working on the problem using AI assistance.
What Happened
The Navier–Stokes problem, selected by the Clay Mathematics Institute in 2000, concerns equations describing fluid flow and whether those equations can break down under certain conditions. On Monday, Buckmaster posted a proof on Mastodon demonstrating that a simplified version of the equations can break down. He and Alpöge had spent nearly a year working on the problem using publicly available models from both OpenAI and Anthropic.
The following day, OpenAI presented a proof showing that the full Navier–Stokes equations can also break down. The company stated that the solution was obtained using an internal model that significantly outperforms its recently released Astra model. OpenAI reported that it does not plan to claim the $1 million prize associated with the solution.
Tensions arose after Buckmaster detailed his interactions with OpenAI employees in a document posted alongside his proof. According to Buckmaster, OpenAI employees presented him with two options: either he and Alpöge post their work first, with OpenAI releasing their solution the next day, or Buckmaster co-author a paper with OpenAI that excludes Alpöge due to his affiliation with Anthropic, a direct competitor. Buckmaster stated that OpenAI employees denied that their agents accessed transcripts of his and Alpöge’s work, but offered no response when asked if the models had been trained on those transcripts.
Why It Matters
This controversy highlights a potential shift in how major mathematical breakthroughs are achieved and attributed. Sébastien Bubeck, a member of the technical staff at OpenAI, acknowledged in a press briefing that the team was inspired to pursue the problem after hearing rumors about Buckmaster and Alpöge’s efforts. While it remains uncertain whether OpenAI’s models directly utilized the researchers' work, the incident raises questions about the transparency of AI-driven research and the impact of corporate rivalries on academic collaboration.
Mark Chen, OpenAI’s chief research officer, reiterated in a press briefing that no agents or employees accessed the transcripts of Buckmaster and Alpöge’s work. However, the article notes that given previous incidents involving the Hugging Face hack, it is unclear if OpenAI has full visibility into the data accessed by its autonomous agents. The situation underscores the tension between the resource-intensive nature of frontier AI research and the traditional norms of open scientific inquiry.
The Bottom Line
OpenAI’s claim to have solved a Millennium Prize Problem marks a significant milestone for AI capabilities, but the accompanying dispute over credit with independent researchers underscores growing friction between corporate AI labs and the academic community. As AI models become essential tools for solving complex mathematical problems, questions regarding data provenance, attribution, and the role of human mathematicians remain unresolved.