OpenAI claims research advances as proof scrutiny continues

OpenAI announced AI-generated research results on October 26, 2026, claiming advances across mathematics, computer science and theoretical physics. The company says its models proved the local smoothing conjecture in every dimension, addressing how waves concentrate and disperse. It also claims to have refuted the Exponential Time Hypothesis, an influential assumption about the computing time required for certain difficult problems. That proposed result does not resolve P versus NP.

In theoretical physics, OpenAI reports a mathematical existence proof for pure Yang-Mills theory. The claim addresses part of a Millennium Prize Problem but leaves its mass-gap requirement unresolved. An earlier AI-generated proposed solution to the Navier-Stokes problem also remains unrecognized as a solution, with the Clay Mathematics Institute continuing to classify the problem as active.

OpenAI released mathematics and computer science research with formal proofs checked using Lean. Such verification can establish logical consistency with stated assumptions, but researchers must still determine whether those assumptions capture the intended problem. Independent experts are examining the claims, while the physics materials have not yet been publicly released. If the results withstand scrutiny, they could expand AI’s role in original research while increasing pressure on human capacity to verify and interpret discoveries.