Three AI labs reportedly shape a frontier-model standards body

The Information reports that Google, OpenAI and Anthropic are assembling an industry-led frontier-AI standards organization tentatively called the Standards Authority for Frontier AI, or SAFA, with a possible late-2026 or early-2027 launch. Discussions reportedly include pre-release model testing, shared safety standards and scientific advisers, with METR founder Beth Barnes and OpenAI Foundation director Paul Christiano among possible names. This remains a reported proposal rather than a joint formal launch; funding, governance independence, publication of test results and consequences for failed models are unconfirmed.