OpenAI released mathematical results from an internal model, some Lean formalizations and reasoning summaries on GitHub while describing plans to support researchers. Mathematicians interviewed by WIRED criticized the presentation format, citations to prior work and the flood of results without time for humans to understand them. AGMAI recommendations likewise emphasize human understanding and responsibility and independent preservation of records, and explicitly state that they do not endorse testing through private models itself. Formal verification and acceptance of results by the academic community are different issues.