16 Mathematicians Sign 'Leiden Declaration' Warning AI Threatens Math Core

Sixteen mathematicians from 15 universities worldwide signed the 'Leiden Declaration,' warning that AI challenges math's reliability, authorship, fairness, and research autonomy. The declaration came from a September 2025 workshop at Leiden University in the Netherlands, with about 60 experts from 10 countries, backed by the International Mathematical Union.

It says math is a human activity for understanding and judgment, not just results. AI is used for writing papers, generating proofs, and peer reviews, but may weaken proof rigor, authorship order, and academic responsibility. It lists five threats: AI proofs can have hidden errors, models may not cite human work, expensive tools widen researcher gaps, and more.

Resource gaps are key. Researchers with advanced AI and high-cost computing may have an edge, while those without it may fall behind. The declaration also worries that AI math results are overhyped, overestimating AI and underestimating human contributions, pushing math toward easy automation and away from deep problems.

It doesn't ban AI but calls for clear rules to keep tools serving math, not weakening its core values.