OpenAI’s Model-Assisted Mathematics Claims Face Verification Questions

OpenAI has publicized hundreds of mathematical results, with reports citing 370 or 372 challenges and a broader release of about 722 papers. Experts question proof completeness, formal verification, provenance, attribution and the burden placed on mathematicians; three results have reportedly been withdrawn. The scale is a notable capability signal, but independent confirmation and the significance of the underlying problems remain unsettled.