OpenAI's AI Solves Millennium Prize Problem

OpenAI announced Tuesday that its AI agents cracked one of the seven Millennium Prize Problems, the kind of open question in mathematics that has stumped researchers for decades. That should have been a clean win for the company. Instead, it's spawned a dispute over credit, involving an NYU mathematician, an Anthropic employee, and questions about whether OpenAI's system built on work it never acknowledged.

Tristan Buckmaster and Levent Alpöge had already been using AI to chip away at the same problem. OpenAI says its models got there independently, though Sébastien Bubeck, a member of the company's technical staff, admitted at a press briefing that his team started after hearing a rumor about what Buckmaster and Alpöge were working on. OpenAI has denied lifting their research.

Whatever actually happened in that specific case, the bigger issue is what it says about where math is headed. AI systems are becoming necessary tools for solving the hardest problems in the field, and the computing power required increasingly sits with a handful of companies that don't operate like universities. That shift raises a harder question: what happens to human mathematicians once the work moves inside corporate labs.