The Association for Human Mathematics, a US group of about 750 mathematicians who refuse to work for commercial AI companies, urged researchers to stop collaborating with OpenAI after it posted 722 manuscripts written by an unreleased internal model. OpenAI put the papers on GitHub, saying they solve or substantially advance hundreds of unsolved problems in math and theoretical computer science, and said it consulted an Institute for Advanced Study advisory group on how to release them. The association countered that the same group had advised AI labs not to run advanced math problems through internal models, warning that it would split the field into labs that pull ahead and everyone else. It called dumping more than 700 papers at once a display of power rather than scholarship.
