OpenAI posts 722 AI maths preprints, mathematicians balk

OpenAI uploaded 722 AI-generated mathematics manuscripts to a GitHub repository on 6 October 2026, claiming progress on 372 open problems in geometry, algebra and computer science. Nature reported that an unreleased OpenAI model produced them.

A day later the company withdrew three manuscripts after a sign error invalidated the argument in one paper and the constructions in two dependents.

The reception from working mathematicians was mostly cold. Alvaro Lozano-Robledo, a number theorist at the University of Connecticut in Storrs, called the volume 'staggering' and asked, 'Why is OpenAI trying to solve hundreds of math problems to be released all at once?' Ursula Martin, a mathematician at the University of Oxford, said OpenAI was 'tossing the community a messy first draft and expecting us to clean it up.' Michael Harris, a number theorist at Columbia University, was shorter: 'I will not be reading these papers.'

Not every reaction was hostile. Levent Alpöge, a mathematician at rival lab Anthropic, posted that whatever one thought of the release, 'it's obviously the most significant moment in mathematical history,' while also noting 'sad stories' for researchers whose projects had been pre-empted. A group of mathematicians put the structural objection in writing: 'Releasing over 700 files at once is not a demonstration of scholarship, but a demonstration of power.'

None of the five remaining Clay Millennium Prize Problems featured in the dump; OpenAI released one Navier-Stokes solution separately in September. An OpenAI spokesperson told Nature that 'we evaluate and develop our models against difficult frontier problems in all sciences.' Three researchers on our tracker had circulated the Nature piece soon after it ran.