OpenAI, Google and Anthropic are building their own AI referee.
The three rivals are reportedly forming a body to test frontier models before they ship. They asked Sriram Krishnan to run it, the adviser who said in June there would be no FDA for AI. No launch is confirmed, and the body could open its doors as late as early 2027.
Why it mattersIf industry writes the safety rules, the people testing the models are paid by their makers. That decides whose word you trust when a lab calls its own model safe.
▶ Related video: AI Regulations & Governance Frameworks Explained
Read in the app — free, in 9 languages
Related stories
Seattle told grocers they can no longer read you to set a price.2026-09-24Sanders wants 20 years in prison for building superintelligent AI.
2026-09-24An OpenAI crawler got around Australia's Medicare blocks.
2026-09-24Senate staff writing AI law cannot use the AI they regulate.
2026-09-23Anthropic's chief asked the UN to slow the AI race down.
2026-09-23