Updated

OpenAI released a large batch of manuscripts—722 papers covering 372 result families—containing solutions to long-standing mathematics problems, produced by an unreleased frontier AI model. The release continues a string of similar disclosures that have impressed many mathematicians while raising concerns about verification, research ethics, and academic conduct. An independent advisory group of elite mathematicians, AGMAI, has been formed to review and vet the claims. The scale and speed of the releases are fueling debate within the mathematical community about how AI-generated proofs should be checked, credited, and integrated into the field.