OpenAI this week released hundreds of claimed solutions to mathematical problems, but its proof methods are said to deviate from the guidelines set by the Advisory Group on Mathematics and AI (AGMAI), convened by the independent Institute for Advanced Study in Princeton, New Jersey. AGMAI released its guiding principles for solving mathematical problems in late September, stating that OpenAI's release explicitly indicates it is evaluating its proprietary models using open research problems, which contradicts AGMAI's primary requirement to "stop testing advanced mathematical problems on proprietary models." Furthermore, out of 719 manuscripts, OpenAI only included model chains of thought in 10 releases, and only 42% of the proofs were formalized, failing to fully adhere to AGMAI's recommendations regarding human understanding and formalization. Some mathematicians have pointed out discrepancies between OpenAI's solutions in natural language and their formalized expressions, raising questions about the reliability of the model's ability to formalize its own solutions without human intervention.