OpenAI Pulls Three AI-Generated Math Papers Over Sign Error

OpenAI withdrew three of the 722 AI-generated math manuscripts it released on October 6, after a sign error in one proof collapsed the argument and the two papers that depended on it.

The flawed paper is 'Algebraicity of Weil classes on split abelian eightfolds.' According to the repository's change history, "a sign error invalidates a stabilization-trace cancellation argument and the construction used by two dependent papers." The dependents pulled with it are 'Algebraicity of Kuga–Satake Correspondences for K3 Surfaces' and 'The rational Hodge conjecture for products of K3 surfaces.' The catalogue now lists 719 manuscripts organized into 372 families.

Alongside the withdrawals, fourteen manuscripts received proof repairs. Four involved 'Lipschitz heights and Ashkin–Teller currents' where authors "repaired crossing, boundary-attachment, conditioning and convergence arguments." Thirteen companion papers were updated to reference the revised editions. Six new formalizations brought the Lean-formalized share to roughly 42 percent, covering 300 of 719 top-line results, Implicator reported.

Named mathematicians urged caution. "Until and unless they release the model and people can replicate their results, I think you should treat any claims about one-shotting problems with a single agent as unverified," MIT's Andrew Sutherland said. University of Toronto's Daniel Litt was blunter: "If we want to know the answers to these math questions, I see no reason why we should ask the company to keep them secret from us."

The change log does not name who caught the sign error, or whether more retractions will follow from the hundreds of unformalized results still in the catalog. It is the first entry on our OpenAI tracker this cycle where the company's own repository, not peer review, flagged a retraction.