OpenAI prepares to release over 100 unsolved math problem solutions

By Splitfeed EditorialLast updated Oct 6, 2026, 17:40 UTC

On October 6, 2026, WIRED reported that OpenAI is preparing to release solutions to more than 100 previously unsolved mathematical problems. The announcement comes alongside notable friction from within the mathematics community, with at least one mathematician quoted by WIRED describing "a perception of mobster behavior" from leading AI companies. The WIRED piece frames the episode as a recurring pattern, noting that this is not the first time OpenAI has generated tension with professional mathematicians.

The stakes here extend beyond any individual set of proofs. Unsolved mathematical problems occupy a particular place in academic culture: they are often the product of years or decades of communal effort, and the question of who gets credit, how results are disclosed, and whether the broader research community is consulted in advance carries real professional weight. If an AI system is generating and publishing solutions to problems that mathematicians have been working on, questions arise about priority, attribution, peer review, and whether the standard norms of mathematical publication are being bypassed entirely. The "mobster" characterization, however colorful, points at something specific: a feeling that power is being exercised in ways that override the conventions a community depends on.

None of the three camps, Pro-AI, Anti-AI, or Middle Ground, have published reactions to this story yet. Pro-AI voices would typically argue that accelerating the pace of mathematical discovery is an unambiguous good, and that concerns about process should not stand in the way of genuine scientific progress. Anti-AI voices would likely focus on the cultural and institutional harm of a single company unilaterally publishing results in a field it did not build, raising concerns about the displacement of human mathematicians and the erosion of academic norms. Middle Ground observers would probably call for OpenAI to establish a disclosure process that involves the mathematics community before publication, treating this as a governance problem with a workable solution rather than a reason to halt the research.

The argument will sharpen considerably once OpenAI makes the actual release and the mathematical community can assess the specific problems involved, the quality of the solutions, and whether any prior coordination with researchers took place. How the broader mathematics community responds, and whether any formal bodies weigh in on attribution or publication standards, will be the clearest signal of how lasting this friction becomes.