Mathematicians have released the Leiden Declaration to oppose the misuse of their work by AI.

      **TL;DR** The Leiden Declaration on Artificial Intelligence and Mathematics, supported by the International Mathematical Union and signed by Fields Medal laureate Peter Scholze, urges mathematicians to address how AI companies are utilizing published research without permission, circumventing peer review, and jeopardizing proof integrity and attribution.

      A group of mathematicians from institutions such as Oxford, Cambridge, ETH Zurich, Columbia, and Northwestern has issued a formal statement urging the mathematical community to confront the risks posed by artificial intelligence to their field. The Leiden Declaration on Artificial Intelligence and Mathematics, released on Monday and endorsed by the International Mathematical Union, represents a significant collective response from a major academic discipline regarding AI companies' use of and, at times, exploitation of published research.

      This 11-page document does not oppose the use of AI in mathematics itself but rather criticisms the methods used by AI companies in handling mathematical work: training models on published papers without authorization, announcing findings via press releases rather than peer-reviewed platforms, undermining proper attribution, and prioritizing commercial interests over the intrinsic value of research. “Mathematics is, and must always be, a fundamentally human endeavor,” stated Ulrike Tillmann, vice president of the IMU.

      **Five Threats to Mathematical Research**

      The Declaration highlights five specific ways in which AI endangers the fundamental values that uphold the credibility of mathematics. First, current AI systems generate arguments that appear plausible but are often unreliable, making it challenging to differentiate them from valid proofs. This issue affects both informal reasoning and formal computer-encoded proofs, complicating the translation between machine and human representations of concepts. The presence of authoritative-looking AI-generated content that contains subtle errors is an issue not limited to mathematics, but in a field that relies on certainty, it poses a grave concern.

      Second, AI models trained on existing mathematical literature fail to appropriately attribute the human contributions that inform their synthesis. The Declaration observes that much of the training data was acquired through “systematic exploitation of licenses and access arrangements not intended for AI use, or outright violations of copyright laws.”

      Third, the growing use of AI is being incentivized in ways that distort hiring practices, funding opportunities, and recognition within the field. Fourth, results are increasingly disseminated via press releases and blogs instead of peer-reviewed journals, aiming for publicity “on market timelines before the typical community evaluation processes in mathematics have the chance to occur.” The Declaration cites Google DeepMind’s AlphaProof, which addressed three International Mathematical Olympiad problems in 2024 but took over a year to present its methods in a peer-reviewed context. Google’s broader AI strategy links mathematical reasoning abilities to indicators of general intelligence, creating commercial pressures to announce findings prematurely.

      Fifth, the independence of mathematics faces threats, with research priorities possibly shifting toward topics more suited to automation, rather than those deemed significant by experts. “Indeed, a broader understanding of the field could be irrevocably lost through the process of automation,” the Declaration cautions.

      **Recommendations**

      The Declaration offers four levels of recommendations. Individual mathematicians are encouraged to disclose all AI tool usage in their papers, maintain personal accountability for result accuracy, refuse AI systems authorship, and “carefully consider which tools to utilize” based on whether their developers uphold the Declaration’s principles.

      Mathematical organizations should ensure that results derived from automated methods comply with standards that address specific risks associated with those methods, safeguard authors’ rights by creating licensing agreements that prevent the use of published works as training data without permission, and insist that findings continue to be published through peer-reviewed channels. While European regulatory frameworks serve as a reference, the Declaration argues that the mathematical community must also independently establish its own standards beyond governmental influence.

      For policymakers, the recommendations are straightforward. “Be wary of the hype,” warns the Declaration. “The technology sector currently has strong commercial incentives to exaggerate the capabilities of their products.” It advocates for enhanced public oversight of the AI industry and investment in public computational infrastructure as alternatives to proprietary systems.

      **Who Signed It**

      The Declaration's significance is enhanced by its signatories, which include Peter Scholze, a Fields Medal recipient and director of the Max Planck Institute for Mathematics, who remarked in a personal statement: “I am contemplating my mathematical ideas without the use of AI and generally avoid reading AI-generated text as much as possible.” Other supporters include Robbert Dijkgraaf, former Dutch minister of education and president-elect of the International Science Council, and Steven Strogatz, a distinguished professor at Cornell focusing on public understanding of science and mathematics.

      Kevin Buzzard, a prominent advocate for formalized mathematics at Imperial College, described it as “a well-thought-out response to the current developments, as AI continues to challenge this field.” The tension between AI capabilities and research integrity highlighted in the Declaration is not exclusive to mathematics; however, mathematicians are among the first academic groups to respond with a coordinated, institutionally supported statement.

      **The Deeper Argument**

      The Declaration’s most challenging section