Mathematicians Urge Collaboration Norms After OpenAI Releases AI-Generated Math Manuscripts

On October 19, 2026, prominent mathematicians including Yau Shing-Tung, Xi Nanhua, He Xuhua, and Tian Ye responded at Tsinghua University to the recent release of AI-generated mathematics manuscripts by OpenAI, saying that AI cannot replace human mathematical thinking and that cooperation norms are urgently needed between the mathematical community and AI companies.

OpenAI had posted 722 manuscripts produced by its internal AI models on GitHub, covering number theory, geometry, algebra, and theoretical computer science, and including approaches to several long-standing open problems. One day later, the company withdrew three manuscripts due to errors, updating the list to 719. The release sparked public debate about the future role of mathematicians.

Yau, a professor at Tsinghua and director of the Qiushi College, said AI is a new tool for exploring truth and should be used freely. Xi, an academician of the Chinese Academy of Sciences, said OpenAI is not a member of the academic community and is not bound by its rules, which has caused considerable trouble for researchers. He Xuhua, vice president of the International Mathematical Union, acknowledged that the released materials contain ideas and results worth absorbing, but stressed that mathematicians and AI companies need to discuss and establish cooperation norms.

The mathematicians emphasized that AI can only explore existing problems and cannot replace human thinking, and that a mechanism for collaboration between mathematicians and AI still needs to be developed.