OpenAI posts 700 AI math preprints, mathematicians recoil

On October 6, OpenAI posted more than 700 AI-generated mathematical preprints to GitHub, written by an unreleased model it calls Astra, Nature reports. The batch does not include a solution to any of the five remaining Millennium Prize Problems. The reaction from working mathematicians has been loud and overwhelmingly hostile.

"Why is OpenAI trying to solve hundreds of math problems to be released all at once?" asked Alvaro Lozano-Robledo, a number theorist at the University of Connecticut, who called the scale "staggering." Michael Harris, a number theorist at Columbia University, told Nature: "I can't bring myself to spend my time studying [these preprints], so I will not be reading these papers."

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 for them."

The alarm reached the field's most senior figures. Terence Tao said he was "deeply frustrated." Fields medallist Hugo Duminil-Copin said the "shock is immense. I am paralysed." Fields medallist Peter Scholze urged restraint, writing that "we are all in this together; mathematics is a marathon, not a sprint." Matt Zaremsky put it less diplomatically in quotes compiled by The Decoder: "a trillion-dollar company showed up and just blasted the whole thing with TNT."

An OpenAI spokesperson said the company evaluates its models "against difficult frontier problems in all sciences" to build better tools for researchers. Not every mathematician agreed with the backlash. Levent Alpöge, a mathematician at Anthropic, acknowledged "sad stories" for researchers whose projects had been preempted but called the demonstration "the most significant moment in mathematical history." Three of the people we track in our Who's Who directory had posted the Nature piece this week.