Mathematicians warn of AI threats to profession as industry encroaches

A group of mathematicians is raising concerns about the increasing influence of technology in their profession, particularly due to the advancements in artificial intelligence. This alert comes shortly after OpenAI showcased an AI model that claimed to have disproved a long-standing mathematical conjecture dating back 80 years. The warning was formalized in the Leiden Declaration on Artificial Intelligence and Mathematics, a document crafted by a team of 16 researchers over the span of eight months, following a conference held in September 2025 at Leiden University in the Netherlands. Released on June 2, 2026, the declaration has garnered support from the International Mathematical Union, which oversees major mathematics conferences and prestigious awards like the Fields Medal. Kevin Buzzard, a mathematician from Imperial College London, emphasized the significance of this shift in interest from tech companies towards mathematical research. He stated, "The Leiden Declaration is a well-thought-out response to the ongoing disruptions caused by AI in our field." The declaration, which has attracted hundreds of endorsements, warns that recent AI innovations threaten the core principles of mathematical research. It particularly notes that these advancements disproportionately impact students and early-career mathematicians, raising concerns about the future of the discipline. One of the key issues highlighted is that AI systems can generate plausible yet unreliable mathematical arguments, making it challenging to differentiate between valid proofs and incorrect ones. This situation places additional pressure on reviewers and risks undermining traditional standards regarding correctness, transparency, and verifiability in mathematical proofs. Leslie Ann Goldberg, head of computer science at the University of Oxford, cautioned that the ease of producing inaccurate AI-generated drafts could overwhelm the academic literature with erroneous claims. She warned, "Once these errors infiltrate the research, they may propagate as new findings are built on these faulty foundations."