OpenAI Achieves Three New Scientific Results in Math and Physics

On October 26, 2026, OpenAI announced three discoveries in mathematics, computer science, and physics. The results include a proof of the local smoothing conjecture in harmonic analysis across all spatial dimensions, a deterministic algorithm that disproves the Exponential Time Hypothesis, and a finding in quantum field theory. This announcement suggests a shift in how original scientific knowledge is generated. By resolving these long-standing conjectures, the findings provide new directions for study in their respective fields.

The history behind these results helps clarify their importance. The local smoothing conjecture, which Sogge proposed in 1991, concerns the behavior of waves when they are focused and suggests that extreme focusing is short-lived. This conjecture is linked to other mathematical problems, including the Kakeya and Stein conjectures. Researchers previously proved parts of these theories in lower dimensions in 2019 and 2025, but the OpenAI model has now produced a proof that holds in every dimension. The creation of a new auxiliary partial differential equation was a central part of this work.

Implementation of the research involved working with external experts who acted as shepherds for each result. OpenAI consulted these specialists to decide on the best release strategies for the data. For the mathematical discovery, the AI output was released publicly to allow the research community to explore its implications independently. The computer science result involved a group of collaborators who worked to explain the findings while keeping the AI output accessible for review. The physics result is still being prepared, with a collaborating group finishing their write-up before the materials are released. This method is designed to ensure transparency and recognition for those involved.

These achievements may lead to wider benefits by encouraging cooperation between AI and human researchers. OpenAI aims to use these collaborations to improve the way complex scientific concepts are understood and applied. Such results could eventually affect technology, education, and research by introducing new insights and methods. A commitment to transparency within the academic community may also help bridge the gap between AI-generated data and traditional research standards.

OpenAI plans to refine its approach based on feedback from the academic community. Ongoing work with experts will help identify the most effective ways to share research findings in the future. As AI becomes more involved in scientific discovery, these results may influence how research is conducted and understood globally. Future work will focus on the implications of these discoveries and continued efforts to advance scientific knowledge.