OpenAI Restructures Safety Committee Led by Zico Kolter to Enhance Transparency and Model Review

According to a recent report by AXIOS, OpenAI has announced significant changes to its board’s safety and security committee. On September 17, it was disclosed that CEO Sam Altman will step down from this committee, which will henceforth consist exclusively of independent board members. Zico Kolter is designated as the chair of this committee. The committee will be briefed on major models, and along with the full board of directors, holds the authority to delay the launch of new models. The committee’s first task was reviewing the new o1 “Strawberry” model, which received a “medium risk” rating in OpenAI’s internal assessments. OpenAI is committed to implementing the committee’s recommendations to enhance transparency, foster collaboration with external entities, and standardize safety practices across various product teams.