OpenAI Solves Navier-Stokes in 4 Days, Triggering Math Revolt by 25 Fields Medalists
Updated
Updated · The Atlantic · Sep 15
OpenAI Solves Navier-Stokes in 4 Days, Triggering Math Revolt by 25 Fields Medalists
3 articles · Updated · The Atlantic · Sep 15
Summary
OpenAI said a swarm of 10,000 AI agents solved the Navier-Stokes problem in under four days, cracking one of mathematics’ seven Millennium Prize challenges after decades of failed human efforts.
The result jolted mathematicians because the 166-page proof is described as nearly incomprehensible, raising fears that AI can now produce landmark results faster than humans can absorb or build on them.
Tristan Buckmaster of NYU alleged OpenAI scooped work he and Anthropic researcher Levent Alpöge had advanced with AI tools, though OpenAI said it pursued Navier-Stokes after seeing promising partial results and denied using his inputs.
On Friday, 25 Fields Medal winners warned that AI companies’ race to solve headline-grabbing problems is misaligned with mathematics, which values not just answers but methods that generate reusable ideas and human understanding.
The dispute has widened into an existential debate over how to reward mathematical work as AI accelerates—from basic arithmetic to elite proofs in a few years—forcing the field to rethink credit, communication and verification.
Could OpenAI's historic Navier-Stokes breakthrough be plagiarized AI slop that threatens mathematical research?
If an AI solves a century-old problem but no human can understand the proof, did it actually advance science?
The $10M Navier-Stokes Proof: How OpenAI’s AI Swarm Sparked a Mathematical and Academic Uprising
Overview
In September 2026, OpenAI rapidly mobilized a swarm of 10,000 AI agents to solve the Navier-Stokes problem after hearing rumors that competitors were close to major mathematical breakthroughs. This effort, sparked by news of progress from mathematicians Buckmaster and Alpöge, led to a public announcement just hours after Buckmaster shared his own results. The move triggered accusations of academic misconduct and concerns about AI models being trained on researchers' unpublished work. As AI-generated proofs became cheap and fast to produce, academic journals faced a flood of questionable submissions, while traditional mathematics departments struggled with shrinking funding as resources shifted toward AI-driven research.