Anthropic, Google, Microsoft, and OpenAI join forces to form AI safety group

Leading AI companies unite to prioritize secure and careful AI development with the establishment of the Frontier Model Forum.

The forum is designed to address the growing security risks associated with the use of large-scale machine learning models, so-called frontier models, which are beyond current capabilities. The joint effort was initiated by Anthropic, Google, Microsoft and OpenAI, which recognizes the need for self-regulation in AI development until regulations are in place.

The Forum plans to focus on four core pillars:

• advancing AI safety research,
• defining best practices,
• collaborating with policymakers, academics, civil society, and companies,
• promoting AI development to address society's greatest challenges.

With a clear commitment to ensuring safety, membership in the Frontier Model Forum requires companies to produce frontier models and demonstrate a dedication to implementing safety measures.

This initiative follows a recent safety agreement between the White House and major AI companies, including the participants in the Frontier Model Forum.