OpenAI has formed a new advisory group of prominent mathematicians to advise it and other leading AI labs on reviewing and communicating mathematical results. The group says it is independent, but its launch caused confusion about its relationship with OpenAI, renewing concerns that the company has mishandled its engagement with mathematicians.

Martin Hairer, a member of the group and a professor at Imperial College London and EPFL, said OpenAI’s approach helped prompt its formation, but the group receives no support from the company and members are not bound by agreements beyond standard confidentiality requirements. He described the launch period as frustrating and intense, and said the group is open to working with other AI labs. Its first major challenge is helping coordinate the release of numerous results from an unreleased OpenAI model, which researchers fear could disrupt their field.