Mathematicians Issue ‘Leiden Declaration’ Against AI Misuse An international coalition of mathematicians has escalated its confrontation with the tech industry, issuing a formal declaration that frames artificial intelligence not just as a powerful tool, but as a potential threat to the foundations of mathematical research.

From Leiden conference to global declaration

The Leiden Declaration on Artificial Intelligence and Mathematics traces back to a conference at Leiden University in the Netherlands in September 2025, where concerns about AI’s rapid advance in mathematics were first systematized by a working group of 16 researchers over eight months. Their worries intensified after OpenAI publicized an AI model it said had disproved an 80‑year‑old conjecture in geometry via press release rather than traditional peer review.

On June 2, 2026, the group released the 11‑page declaration, endorsed by the International Mathematical Union (IMU) and already signed by hundreds of mathematicians, including Fields Medalist Peter Scholze. The IMU, which oversees major prizes like the Fields Medal, lent institutional weight to the message. Kevin Buzzard of Imperial College London called the declaration “a well-thought-through response to what is currently happening, as AI continues to disrupt this space.”

Core grievances and ethical concerns

The document does not reject AI outright but condemns how AI companies are using mathematical work: training models on published papers without consent, bypassing peer review through splashy announcements, and reshaping research priorities around commercial value rather than intellectual significance. It warns that current AI systems can “produce plausible but unreliable (or even incorrect) arguments” that are hard to distinguish from real proofs, jeopardizing “correctness, transparency, and independent verifiability.”

The declaration also accuses companies of “systematically exploiting licences and access arrangements that were not made with artificial intelligence in mind, or indeed by simply violating copyright protections,” and failing to properly attribute the human work embedded in their models. This, mathematicians argue, threatens students and early‑career researchers most, and risks cluttering the literature with cheap, inaccurate AI‑generated drafts that can propagate errors through future work.

A call to defend mathematics as a human endeavour

Signatories outline five categories of threats, from unreliable proofs to distorted incentives in hiring and funding, and propose individual, institutional, and policy responses to protect the field’s integrity. IMU vice president Ulrike Tillmann summarized the stakes succinctly: “Mathematics is, and should always remain, a profoundly human endeavour,” even as AI tools grow more capable.

Continue reading https://foxvector.com/stories/019e8a9d-c2d9-20a0-729c-33e6c97235c9