OpenAI, the company behind the powerful AI system GPT, has created an independent advisory group composed of top mathematicians to help guide the release and communication of recent mathematical results it has achieved. This group, known as the Advisory Group on Mathematics and Artificial Intelligence (AGMAI), will offer guidance on how to present and share these results. The group is hosted by the Institute for Advanced Study (IAS), a prestigious research institution in Princeton, New Jersey, known for its contributions to mathematics and theoretical physics. Its nine members come from leading universities like Stanford, Harvard, Oxford, and Cambridge, and include individuals who have won prestigious awards such as the Fields Medal and MacArthur "genius grants." The advisory group operates independently of OpenAI, with the freedom to offer advice even if it wasn’t requested, to publicly comment on the impact of its work, and to share its recommendations with the public. Members are not paid by OpenAI, and the group can change its composition as needed. OpenAI has granted the group early access to its research to ensure they can provide informed guidance. The group was formed after OpenAI approached some mathematicians about creating an external advisory board. However, after discussions with the company, the mathematicians decided to establish their own independent group instead. It remains unclear which members were initially contacted by OpenAI or how the rest were selected. In a blog post, group member Martin Hairer explained the group’s formation, addressing concerns about misinformation surrounding its creation. He emphasized that the group is truly independent and that members have not signed any agreements that would limit their ability to speak publicly. They rely on the IAS for technical support. Hairer noted that the group is not a representative sample of the broader mathematical community, but all members are committed to their role and acting in good faith. However, the group’s small size has raised questions about whether it has the range of expertise needed to evaluate the wide variety of mathematical results that may come its way. Some mathematicians have expressed skepticism about whether the group truly represents the community affected by AI’s growing influence on mathematical research. Whether the group can gain the trust of mathematicians will be key to its effectiveness in addressing concerns about how OpenAI has handled the release of its recent results. Hairer acknowledged that AI companies may attempt to reinterpret the group’s advice to align with their public relations goals. The group has yet to decide on specific recommendations but is focused on addressing a clear issue: AI companies, including OpenAI, have recently produced notable mathematical results, but their publication and communication have not met traditional standards of academic rigor. OpenAI has claimed that its new, unreleased AI model has solved over 100 long-standing mathematical problems, and the advisory group’s first task is to help coordinate the responsible release of these results.