OpenAI withdrew three mathematical manuscripts from its public repository the day after publishing 722 papers generated by an internal frontier model. The withdrawal, recorded in the repository’s history.md file on October 7, was due to a sign error that invalidated an argument in one paper and the construction used by two dependent papers.
The three withdrawn manuscripts are: “Algebraicity of Weil classes on split abelian eightfolds”, “Algebraicity of Kuga–Satake Correspondences for K3 Surfaces” and “The rational Hodge conjecture for products of K3 surfaces”. Each now carries notices explaining the gap.
In addition to the withdrawals, OpenAI revised 14 other manuscripts with proof repairs, corrected statements and clarified hypotheses. The catalog went from 722 to 719 manuscripts, still organized into 372 result families. About 42% of the top-line results have Lean formalizations, according to repository updates.
The original publication on October 6 was presented by OpenAI as an effort to advance mathematical knowledge with AI-produced results. The company said the process is iterative and that it continues to update the repository with new formalizations and errata. A spokesperson said the advisory group recommended releasing without waiting for full formalization and that roughly half the results went out without full confirmation.
This update complements the Geekiki’s previous coverage of the initial release of the papers.
Sources
This content was produced with artificial intelligence assistance and reviewed by GeekikiBot.
Por GeekikiBot
