OpenAI announced in early October 2026 that its AI systems had solved numerous open mathematical problems across fields including group theory and topology. The announcement prompted varied reactions from mathematicians worldwide, documented in a blog post by Proofs and Prompts that compiled over 100 responses. These reactions capture the community's initial response to what many described as a seismic development in mathematical research.
Many mathematicians expressed astonishment at the technical capabilities demonstrated. Xiaolei Wu, a mathematician working in group theory, reported being 'surprised to find many messages containing the same PDF file' detailing solutions to 'almost all the major open problems' in their field, including the non-amenability of Thompson's group F and the falsity of the Eilenberg–Ganea conjecture. Roman Sauer noted that a mathematical technique they had developed - replacing a manifold with a measurable foliated object from ergodic theory - appeared in OpenAI's solutions to specific problems (207 and 335), describing the effectiveness of this approach as 'unreasonable' and stating it had surprised them 'a few times' before seeing it in the AI-generated work.
The reactions also revealed concerns about how AI-generated work aligns with established research practices. Nicholas Williams emphasized that mathematicians' work extends beyond solving open problems to 'grappl[ing] with the subject matter and think[ing] about what the good conjectures should be in the first place,' suggesting a broader role for the discipline that may be overlooked in assessments focused solely on problem-solving speed. Henry Wilton criticized OpenAI's approach on multiple fronts, noting that by not attributing authors to their papers and avoiding traditional journal submission, the company undermines community norms essential for 'structure, a yardstick for progress and long-term goals,' while also failing to disclose 'fundamental information about how the problems are selected or the failure rates of their attempts.' Wilton concluded that this approach risks washing away 'our community’s structures to further their financial interests,' reflecting a concern that the rapid dissemination of AI solutions could disrupt the collaborative and verificatory processes that have historically defined mathematical advancement.