OpenAI Taps 9 Elite Mathematicians to Vet AI Results as 100-Plus Proof Claims Draw Scrutiny
Updated
Updated · The Verge · Sep 23
OpenAI Taps 9 Elite Mathematicians to Vet AI Results as 100-Plus Proof Claims Draw Scrutiny
3 articles · Updated · The Verge · Sep 23
Summary
Nine mathematicians from institutions including Stanford, Harvard, Oxford and Cambridge will advise OpenAI on reviewing, communicating and releasing AI-generated math results through the new AGMAI panel.
OpenAI says the group is independent, unpaid and free to publish its advice, while member Martin Hairer said it was formed after OpenAI approached some mathematicians amid a "torrent of misinformation."
Researchers called the panel a useful first step but questioned its opaque selection process, limited representativeness and unclear influence over OpenAI's decisions.
The urgency is OpenAI's claim that an unreleased model has solved more than 100 long-standing open problems, after earlier math announcements triggered accusations of poor attribution, weak coordination and PR-driven dissemination.
Could OpenAI's unreleased model have secretly memorized human mathematicians' unpublished work to crack the Navier-Stokes equation?
When machines generate proofs too complex for human minds to grasp, have we solved math or simply made ourselves obsolete?
If an AI solves a Millennium Prize problem in 88 hours, does the machine or the human verifier deserve the reward?
The 2026 Navier-Stokes AI Proof: Verification, Ethics, and the Transformation of Mathematics
Overview
In September 2026, OpenAI responded to rumors of progress by academic researchers on the Navier-Stokes problem by rapidly deploying a powerful internal AI system with thousands of agents. This system generated a groundbreaking proof, which was then formally verified using Lean. The announcement sparked controversy over attribution and data ethics, as some researchers feared their work could be absorbed by commercial AI tools. Leading mathematicians warned that AI could overwhelm human understanding and disrupt the traditional transmission of mathematical knowledge. In response, the global math community formed an independent Math Advisory Group to ensure proper vetting and attribution of AI-generated proofs, aiming to balance rapid AI discovery with human comprehension and trust.