What happens when a mathematical proof is no longer the work of a human, but of an AI model to which academic researchers have no access? Who is responsible for errors, and who receives credit when a proof is correct? These are no longer hypothetical questions, but pressing dilemmas facing mathematicians around the world. To address these challenges, an international group of researchers presented the Leiden Declaration on Artificial Intelligence and Mathematics on 16 June 2026.
The declaration focuses on the fundamental values of mathematics that are coming under pressure due to the growing use of AI in the field. The use of AI raises questions about accountability for errors in proofs, about the fair attribution of scientific credit, and about how to determine whether an AI-generated proof is genuinely new or merely a reformulation of existing work without proper attribution.
The declaration contains concrete recommendations for individual mathematicians, mathematical organisations and funders, policymakers, and commercial AI laboratories. From VU Amsterdam, mathematicians Sander Dahmen and Alain Chavarri Villarello were involved in its development. The declaration is endorsed by the International Mathematical Union (IMU). Vice-president Ulrike Tillmann stated that the IMU takes the rapid development and impact of artificial intelligence on the field very seriously, and that the future of mathematical research must be guided by human judgement, fair and transparent practices, and the shared values of the global mathematical community.
Within a week of its original announcement, the declaration was signed by more than two thousand scientists, including Fields Medal winners Terence Tao, Elon Lindenstrauss, Martin Hairer, Peter Scholze, Hugo Duminil-Copin, James Maynard, and Maryna Viazovska. The declaration attracted international media coverage in outlets including The New York Times, Scientific American, Science, and de Volkskrant, as well as dozens of other national and international news reports.