OpenAI claims it solved the Navier-Stokes existence and smoothness problem, one of mathematics’ seven Millennium Prize problems. The company deployed roughly 10,000 agents and tens of millions of dollars in compute power to crack the code in just 88 hours. In any other era, this would be a cause for universal celebration. Instead, the announcement has triggered a firestorm within the academic community, revealing deep fractures between traditional mathematical norms and the aggressive, resource-heavy tactics of big tech.
The core of the controversy isn’t just about who solved it first, but how. NYU professor Tristan Buckmaster accuses OpenAI of leveraging his prior work via its Codex tool without proper attribution or consent. He describes a tense negotiation where OpenAI researcher Sébastien Bubeck allegedly offered him “unlimited compute” and sole authorship on the breakthrough paper, provided he excluded his collaborator, Levent Alpöge of Anthropic. Buckmaster rejected what he called a “bribe,” choosing instead to go public with his allegations. OpenAI spokesperson Laurance Fauconnet categorically denied that Buckmaster’s prompts influenced their system, stating it was “impossible” for his recent activity to have played a role.
This incident highlights a cultural clash that has been brewing for years. Mathematicians view their field as a collaborative, slow-burn endeavor focused on collective understanding. OpenAI, by contrast, appears driven by a competitive urge to win against rivals like Anthropic. The involvement of Alpöge, a researcher at a direct competitor, seems to have escalated OpenAI’s urgency. While Buckmaster remains skeptical of OpenAI’s denials, noting their past behavior warrants caution, the damage to trust is already done. Colleagues have turned Buckmaster’s office into a “war room,” scrutinizing math and coordinating legal outreach.
The implications extend far beyond fluid dynamics. If AI labs can bypass academic etiquette by sheer force of computational spending, the incentive structure for pure research changes fundamentally. We are witnessing a shift from discovery as a public good to discovery as a proprietary asset in a corporate arms race. The question now isn’t whether AI can solve hard problems, but whether the human cost of these victories will discourage the very experts needed to verify them.
