Twenty-five Fields Medal winners, including Terence Tao, Artur Avila, and Manjul Bhargava, published a joint declaration on September 11, 2026. The mathematicians argue that commercial incentives in the AI industry are creating a fundamental misalignment with the rigorous standards of mathematical research.
The signatories state that AI companies prioritize speed and benchmark performance over the deep verification processes essential to mathematics. This approach leads to premature claims regarding complex problems, including those in the Millennium Prize category, which are later retracted with little public notice.
The declaration highlights that the current cycle of flashy claims attracts funding and talent to AI projects, while the burden of verification remains with the academic community. This document follows the Leiden Declaration of June 2026, which previously addressed risks AI poses to expert professions.
While the declaration carries moral authority, it lacks formal enforcement mechanisms. The authors aim to establish a reference point for policymakers, journal editors, and funding agencies to help evaluate AI-generated mathematical claims and maintain the integrity of the field.
