The International Mathematical Union has officially endorsed a stark warning about artificial intelligence's encroachment into mathematics, backing a declaration that accuses tech companies of operating on "commercial logic" antithetical to mathematical values. Published June 2, 2026, the Leiden Declaration on Artificial Intelligence and Mathematics was developed by a working group of 16 researchers over eight months following a conference at Leiden University in the Netherlands last September. The IMU's endorsement carries significant weight—it oversees the Fields Medal and hosts mathematics' most prestigious gatherings. "Mathematics is, and should always remain, a profoundly human endeavor," said Ulrike Tillmann, vice president of the IMU, in a statement.

AI's Proof Problem

The declaration's sharpest criticism targets AI's tendency to generate plausible but incorrect mathematical proofs that are difficult for reviewers to catch. "Inaccurate AI-generated drafts are cheap to produce, and there is a risk of cluttering the literature with claimed results that are simply wrong," said Leslie Ann Goldberg, head of computer science at Oxford University. She warned that once erroneous results enter the mathematical canon, errors propagate as future research builds on faulty foundations. The declaration warns this dynamic is "jeopardizing our ability to implement traditional standards for the correctness, transparency, and independent verifiability of proof." Early-career mathematicians face disproportionate pressure as AI-generated text floods academic channels.

The OpenAI Flashpoint

The timing of the declaration feels deliberate—it arrives just two weeks after OpenAI publicized one of its models as having disproved an 80-year-old mathematical conjecture in geometry, the same day the company reportedly prepared to offer stock shares to the general public. Mathematicians involved in drafting the document pointedly noted that corporate press releases highlighting AI achievements operate on "market timelines before the accepted processes of community evaluation in mathematics can take place." OpenAI uploaded a research paper with commentary from independent mathematicians but declined to disclose prompts, training data, and computational resources used—basic information needed to assess the scientific meaning of their result. "The AI model is proprietary and unavailable to anyone outside the company," said Rodrigo Ochigame, a historian and anthropologist of computing at Leiden University. "We get a flashy promotional video, while basic information needed to assess the scientific meaning of the result is kept secret."

Copyright and Citation Grievances

Beyond proof verification, the declaration highlights how AI models trained on published works frequently return outputs that fail to cite the human sources they synthesize—creating attribution nightmares for mathematicians trying to give credit where it's due. The document also calls out how many current AI systems were trained on data obtained through "exploiting licenses and access arrangements" or "simply violating copyright protections." This matters: when a proprietary model generates a proof, who owns it? And what happens to the careers of human mathematicians whose work was scraped without consent to train those models in the first place?

Who Gets Left Behind

The declaration warns that AI adoption may become incentivized for its own sake in hiring, funding, and recognition—leaving out researchers unwilling or unable to use technologies "controlled by organizations whose values they do not share." Peter Scholze, director of the Max Planck Institute for Mathematics, offered a pointed analogy on the Leiden Declaration website: "In my experience, mathematical ideas, like children, must be nurtured and grow over the years. Just like I do not want my children to be educated by AI, I am pondering my mathematical ideas without use of AI." Ursula Martin, mathematician and computer scientist at Oxford, noted that OpenAI's achievement was "remarkable" but likely required massive compute resources—resources a human team with equivalent effort probably would have solved the same problems. She cautioned that mathematics is fundamentally about "the cultivation of ideas, understanding, judgment and human insight," not just problem-solving benchmarks.

What Comes Next

The declaration offers concrete recommendations: mathematicians should transparently disclose AI tool usage while retaining responsibility for correctness; professional organizations should develop guidelines for AI in publication and review, protect authors' rights through licensing that prevents unauthorized training data use, and prepare to intervene when major results are claimed through unconventional means. For policymakers, the document is blunt: protect author rights, regulate the AI industry, invest in public computational infrastructure—and "don't believe the hype." The declaration acknowledges the elephant in the room—lucrative tech jobs and computing resources look attractive "in an era of underfunding of higher education and precarious academic employment"—but insists any collaborations between mathematicians and tech companies must abide by its standards. Kevin Buzzard at Imperial College London put it simply: "Mathematicians should find it quite striking that tech companies are suddenly interested in their work."

Key Takeaways

  • The Leiden Declaration warns AI-generated proofs are cheap to produce but may be wrong, risking pollution of the mathematical literature with errors built on faulty foundations
  • OpenAI's recent claim to disprove an 80-year-old conjecture highlights transparency concerns: proprietary models, undisclosed training data, and promotional timelines that bypass peer review
  • Mathematicians face pressure to collaborate with tech companies amid academic underfunding, raising questions about who controls the discipline's future direction

The Bottom Line

This isn't just an academic squabble—it's a proxy war over whether mathematics serves human understanding or becomes another product category for Silicon Valley. When Peter Scholze compares AI assistance to letting strangers raise your kids, you know the community feels genuinely threatened. The IMU backing this declaration signals mathematicians aren't going quietly into that algorithmic night.