Why Mathematicians Are Sounding the Alarm on AI’s Growing Reach

A group of prominent mathematicians has issued a stark warning about artificial intelligence’s impact on their field, releasing a declaration that calls attention to the risks AI poses to research integrity and the future of the discipline. The Leiden Declaration on Artificial Intelligence and Mathematics, published June 2, 2026, comes just weeks after OpenAI claimed one of its models had disproven an 80-year-old geometric conjecture—a move that caught many in the academic community off guard. Drafted over eight months by 16 researchers following a September 2025 conference at Leiden University, the declaration has already earned backing from the International Mathematical Union, the organization behind the Fields Medal. It warns that AI tools are eroding the “characteristic values” of mathematical work, particularly for students and early-career researchers. “Tech companies suddenly caring about mathematics should raise eyebrows,” said Kevin Buzzard, a mathematician at Imperial College London. “This declaration is a measured response to real disruption.” The document flags two core concerns. First, AI models can generate arguments that look convincing but are subtly wrong, making peer review harder and threatening standards of proof. Second, low-cost AI-generated drafts risk flooding journals with faulty claims. “When errors get published, they become foundations for more errors,” warned Leslie Ann Goldberg, head of computer science at the University of Oxford. With hundreds of signatories already, the declaration signals a profession grappling with how to preserve rigor in an era of automated output.