Less than a month after OpenAI claimed to have cracked a major problem in mathematics, sparking fierce debate within the math community, the artificial intelligence company has released more than 700 new research papers, unveiling results across multiple areas of mathematics.

On Tuesday, OpenAI said these new results were generated by a model that has not yet been publicly released, and that this model outperforms commercially available versions. The company posted 722 papers on the code-hosting platform GitHub, containing solutions to hundreds of previously unsolved mathematical problems.

OpenAI said the approach partly followed recommendations from an independent body, the Mathematics and AI Advisory Group. Last month, after OpenAI announced it had cracked the notoriously difficult "Navier-Stokes existence and smoothness problem," it said it would begin collaborating with the advisory group.

The Navier-Stokes results triggered enormous controversy: on one hand, some questioned whether OpenAI's model gained an advantage from data generated by human mathematicians using its products; on the other hand, prominent mathematicians argued that AI-generated solutions of this kind are harming the discipline of mathematics.

On Tuesday, the advisory group issued a statement on its website acknowledging OpenAI's involvement while also sounding a warning. "The future of mathematical research cannot simply be interpreting results produced by AI labs," the statement read. "Mathematicians must be able to pose their own questions, build their own research approaches, and explore directions that have not been selected as demonstrations of AI capability."