Summary
ColdFusion examines the controversy around OpenAI’s claim to have made major progress on more than 300 unsolved mathematics problems, including a result on the Navier-Stokes Millennium Prize problem. The video opens by arguing for nuance in how we judge AI, acknowledging both the industry’s real problems and its genuine strengths in areas like coding and high-level mathematics.
The story centers on NYU mathematician Tristan Buckmaster, who says OpenAI began work on the problem only after learning of his own progress, and used the same niche approach he had pursued for a year. The video lays out the timeline of conversations with OpenAI’s Sebastien Bubeck, the accusations of stolen work and intimidation, and OpenAI’s own account of how it proceeded.
According to the video, OpenAI used a multi-agent system running about 10,000 agents for roughly 88 hours, exchanging 2.7 million messages and consuming around 130 billion output tokens, with GPT-6 Astra then spending 17 hours formalizing the proof. The video explains what a finite-time blow-up means for fluid equations and asks who deserves credit when AI contributes to a breakthrough, making it a good overview of the trust and attribution questions raised by AI-driven mathematics.
📺 Source: ColdFusion · Published October 08, 2026
🏷️ Format: News Analysis
