More Than 1,000 Mathematicians Pledge to Keep AI Out of Their Research

More than 1,000 mathematicians have now signed pledges to keep AI companies out of their research, through a new group called the Association for Human Mathematics (AHM). The movement has grown quickly since OpenAI posted 722 AI-generated math manuscripts in a single batch on Oct. 6, 2026, a release the group calls “a demonstration of power” rather than scholarship. The pledge drive was trending widely online on Tuesday, Oct. 27.

What the Association for Human Mathematics asks

The AHM was founded in September, German tech site heise reported, and had listed more than 800 members at the time of its October statement, including Fields Medal winner Peter Scholze of the Max Planck Institute for Mathematics. By Oct. 20, membership had passed 1,000, including professors and Ph.D. students, according to New Scientist.

Members are asked to make up to three pledges, and the third is optional. New Scientist described two resulting tiers. Members nicknamed “vegetarians” agree not to work with or for AI companies and not to use chatbots to write papers, lecture notes or books. “Vegans” go further and avoid AI in their research entirely.

Sam Raskin of Yale University told New Scientist he signed the first two pledges but not the third, because it was unclear whether his occasional use of AI for simple tasks would count. He called the pledges “pretty mild” and said members join for reasons ranging from environmental concerns to job security. Raskin also criticized what he described as community leaders “quietly being paid by either Anthropic or OpenAI” without always disclosing it. “I think that’s toxic,” he said.

OpenAI’s 722 math papers

The backlash traces to OpenAI’s Oct. 6 release, which Science magazine said covered 372 distinct results. OpenAI told Science it used an unreleased ChatGPT model to attempt about 4,000 problems, and that most solutions took the equivalent of about three hours of ChatGPT Pro “thinking.” That followed a September preprint in which OpenAI said it had solved the Navier-Stokes problem, one of the Clay Mathematics Institute’s Millennium Prize Problems, Nature reported.

The AHM responded the next day. “Mathematicians did not ask for this work to be done,” its statement said, arguing that OpenAI ignored the central advice of the independent Advisory Group on Mathematics and Artificial Intelligence, which had said frontier AI companies should not test advanced problems on internal models. “Releasing over 700 files at once is not a demonstration of scholarship, but a demonstration of power,” the group wrote, urging mathematicians “to discontinue their work with OpenAI.”

Many mathematicians say the work is impressive but hard to use. Edward Witten, a mathematical physicist at the Institute for Advanced Study, told Science that a human who understood the results “could write them in a more understandable way.” Steven Sivek of Imperial College London said one paper close to his own field ran 140 pages and was “completely unreadable.” Science also reported that OpenAI had withdrawn three manuscripts and revised 14, and that only about 300 of the remaining 719 had been formalized in Lean, a programming language that lets a computer check each step of a proof.

‘We’re not denialists’

Sivek, a “vegan” member, told New Scientist the group is not just a list of AI opponents. It has set up committees on protecting preprints from being swept into AI training data, on career problems facing professors and on how undergraduate and graduate teaching must change. “We’re not denialists. We’re not sticking our heads in the sand and saying ‘this is not happening,'” he said. “It is happening, it’s real. We need to figure out how to adapt the profession to this new emerging reality.”

For others, the objection is about the craft itself. “I’m not comfortable putting out things that I don’t understand, that I can’t vouch for,” Sivek said. Hervé Le Dret of Sorbonne University cited ethical concerns including what he called AI’s “scandalous waste of resources and environmental impact.” UCLA’s Terence Tao warned earlier that the way AI results are being “dumped” on the field could harm it, New Scientist noted.

OpenAI’s response

OpenAI did not respond to New Scientist’s request for comment. A spokesperson told Nature that the company’s goal is to provide tools for mathematicians and scientists, and that “we evaluate and develop our models against difficult frontier problems in all sciences.” OpenAI told Science it is “extremely supportive of mathematicians working on our results,” and said it plans to fund workshops and is talking with potential partner institutions.

The fight is part of a wider argument about where human work fits as AI improves. For related coverage, see Shigeru Miyamoto’s view that humans will always have a role in game design, OpenAI’s plan to train its most powerful models without internet access and Bill Gates’ forecast of 20 tumultuous years.

Sources

• New Scientist: 1000 human mathematicians pledge to shun AI in their research
• Association for Human Mathematics: Statement on OpenAI’s October 6 release
• Science: OpenAI just changed math forever. Exactly how, no one yet knows
• Nature: OpenAI posts 700 maths preprints online
• heise online: AI-generated math solutions draw sharp criticism