# Terence Tao Criticizes OpenAI’s Release of AI-Generated Math Papers

By Simon Yoon

Canonical URL: https://www.tokenpost.com/news/technology/34312
Published: 2026-10-18T09:58:47.000Z
Updated: 2026-10-18T09:58:47.000Z
Section: Technology

> The collection includes 719 manuscripts in 372 result families. Tao called one result tied to the Riemann Hypothesis the field’s largest advance on the problem in 50 years.

Mathematician Terence Tao criticized OpenAI’s mass release of AI-generated mathematical manuscripts, saying researchers already working on the problems were not notified before the release and some junior researchers’ projects were abruptly overtaken.

In an interview published Thursday, Oct. 15, Tao described the lack of notice as deeply discourteous. He said he had reviewed about 10 papers. Some extended existing research, one or two offered clever new ideas, and many remained difficult for mathematicians to understand.

The collection lists 719 manuscripts grouped into 372 result families. An unreleased internal model was given about 4,000 open problems. The vast majority of results used the standard process, which averaged three hours of ChatGPT Pro thinking compute per result. The zero-free-region work involving the Riemann zeta function was an exception. About 42% of top-line results have Lean formalizations; some results without formalizations may have problems.

Tao said moderately difficult problems have traditionally helped students and postdoctoral researchers build skills and establish a record. He worries that a wave of AI-generated solutions could make mathematicians less willing to share questions they hope to pursue, for fear that a system might solve them first.

One result claims a zero-free region for Dirichlet L-functions, including the Riemann zeta function, when the real part exceeds 7/8. The Riemann Hypothesis says all nontrivial zeros of the zeta function lie on the line at real part 1/2. Tao called the 7/8 result the largest advance on the problem in 50 years, but stressed that it does not prove the hypothesis. He said another major idea would be needed to move the result beyond a real part of 3/4.

Tao also argued that the race among AI companies to release stronger systems has come at the expense of human understanding and safety. He warned that deploying systems in real-world settings without understanding how they work could eventually cause real problems.

He expects mathematics to move toward larger, more industrialized collaborations, while arguing that traditional approaches should remain part of education, much like teaching children to calculate by hand before using calculators.