OpenAI announced on Tuesday that its internal artificial intelligence system had produced solutions to more than 300 previously unsolved mathematical problems across diverse fields such as algebra, geometry and differential equations. The results, which had eluded human researchers, were shared with limited publicity.

The announcement follows a month of criticism directed at OpenAI after it disclosed a solution to a modified version of the Riemann hypothesis, one of several Millennium Prize problems carrying a $1 million reward. While the prize problems themselves remain unsolved, the company highlighted its progress on a related variant.

Mathematicians have expressed mixed reactions to the development. Alex Kontorovich, a professor at Rutgers University, noted during a gathering at the National Museum of Mathematics that the achievement marked a shift in expectations within the field. "Among analytic number theorists, we would discuss, ‘isn’t it embarrassing that we still can’t get anything nearly of that strength,’" he said.

The release has reignited debate over the use of proprietary AI models to evaluate mathematical capability. An independent advisory board that OpenAI consulted previously urged the company to cease testing advanced mathematical problems on closed systems, stating that such practices do not align with the goal of advancing human knowledge. OpenAI responded by emphasizing its commitment to ensuring that progress in artificial intelligence contributes to broader scientific discovery.

The company has not indicated whether further mathematical results will be made publicly available.