Home / Technology / Mathematician Accuses AI of Stealing Math Breakthrough
Mathematician Accuses AI of Stealing Math Breakthrough
19 Sep
Summary
- Mathematician claims OpenAI's AI used his work to solve a legendary math problem.
- OpenAI's AI models are useful, making it hard for mathematicians to avoid them.
- AI's rapid advancements challenge traditional attribution and understanding in math.

Mathematician Tristan Buckmaster claims OpenAI's AI copied his approach to solve the legendary Navier-Stokes existence and smoothness problem, which carries a $1 million bounty. He stated that OpenAI deployed tens of thousands of AI agents to reach the solution only after learning the equation was close to being solved. Buckmaster, while accusing OpenAI, admits to using AI tools like Codex and Claude for his own research due to their utility and the limited choices available.
This incident has sparked debate about AI's impact on mathematics and whether it could render human mathematicians obsolete. OpenAI investigated the claims and stated that Buckmaster's prompts from two months prior to their September 8, 2026 announcement could not have influenced the system. Similarly, mathematician Andreas Thom expressed surprise when OpenAI's Astra model used techniques he developed, questioning how the AI learned them.
Mathematicians like Alex Townsend at Cornell University express both excitement and nervousness about AI's capabilities, pondering their own purpose in a field increasingly influenced by AI. While some mathematicians, like Thom, have adopted AI tools with privacy settings, others are more critical. A group of 25 Fields medalists and over 4,000 signatories of the Leiden Declaration have voiced concerns about AI companies and mathematicians being misaligned.
Buckmaster advocates for a truce, proposing that mathematicians and AI laboratories establish ground rules for releasing results and ensuring proper attribution. He plans to revise his own prematurely published papers, calling them "AI slop." Despite attempts to slow down AI's integration, the efficiency gains offered by AI suggest its continued adoption, potentially isolating early-career researchers who do not utilize these tools.