Google, OpenAI, and Anthropic plan joint AI safety body SAFA by early 2027

Google, OpenAI and Anthropic are reportedly creating an independent AI safety body called SAFA, with a possible launch by early 2027 and former White House adviser Sriram Krishnan as CEO. The Standards Authority for Frontier AI would operate independently of government control, turning the companies' safety commitments into testable benchmarks and helping developers evaluate advanced models before commercial release. Arati Prabhakar is another CEO candidate, and the firms are debating whether SAFA should conduct its own model testing. The move follows an OpenAI confirmation of talks with rivals about a shared safety framework, and a July proposal by Google DeepMind's Demis Hassabis for an industry self-regulator. It comes amid pressure after Anthropic CEO Dario Amodei urged slower rollout of frontier models.