Google, OpenAI and Anthropic are planning an independent AI safety body that could launch by late 2026 or early 2027. The group is provisionally named the Standards Authority for Frontier AI, or SAFA. It would operate without government control and set practical benchmarks for the safety commitments the three companies have already made public. It could also support developers testing advanced models before commercial release and may conduct its own assessments. The companies have approached former White House AI policy adviser Sriram Krishnan to serve as chief executive and have also considered Arati Prabhakar. The move follows recent calls from Anthropic’s Dario Amodei, OpenAI’s Sam Altman and Google DeepMind’s Demis Hassabis for stronger safety frameworks. Meta and Nvidia have publicly opposed industry-wide coordination, preferring each company to handle its own safety.
Google, OpenAI and Anthropic are planning an independent AI safety body that could launch by late 2026 or early 2027. The group is provisionally named the Standards Authority for Frontier AI, or SAFA. It would operate without government control and set practical benchmarks for the safety commitments the three companies have already made public. It could also support developers testing advanced models before commercial release and may conduct its own assessments. The companies have approached former White House AI policy adviser Sriram Krishnan to serve as chief executive and have also considered Arati Prabhakar. The move follows recent calls from Anthropic’s Dario Amodei, OpenAI’s Sam Altman and Google DeepMind’s Demis Hassabis for stronger safety frameworks. Meta and Nvidia have publicly opposed industry-wide coordination, preferring each company to handle its own safety.