OpenAI announced the formation of the Frontier Model Forum on 26 July 2023, a new industry body focused on AI safety and responsible development of frontier models [source]. The Forum's founding members include Anthropic, Google, Microsoft, and OpenAI.

The organisation states it will focus on advancing AI safety research, identifying best practices for frontier model development and deployment, facilitating information sharing among policymakers and industry, and supporting efforts to develop applications addressing society's greatest challenges. The Forum plans to establish an advisory board drawing from civil society, governments, academia, and the private sector.

OpenAI's announcement describes frontier models as "large-scale machine learning models that exceed the capabilities currently present in the most advanced existing models" and notes these systems "pose increasingly significant risks." The Forum commits to developing and deploying frontier models safely, though specific technical commitments or enforcement mechanisms were not detailed in the announcement.

The timing follows increased regulatory attention on AI systems in multiple jurisdictions. The announcement states the Forum will collaborate with policymakers, academics, civil society, and companies to promote responsible AI development, but does not specify binding obligations for member companies beyond voluntary participation in working groups.

The Forum's structure includes plans for a dedicated fund to support third-party AI safety research and the development of safety standards. Member companies stated they will publish safety frameworks, though publication timelines were not provided. The announcement indicates membership is open to organisations demonstrating a "strong commitment to frontier AI safety" that meet forthcoming membership criteria.