Google, OpenAI, and Anthropic form joint AI safety body

Google, OpenAI and Anthropic are forming a joint safety body tentatively named the "Frontier AI Standard Agency" to establish standardized regulations, filling a gap in government oversight amid growing AI risk concerns. The independent agency follows negotiations between OpenAI and Anthropic for a legally binding agreement to test each other's commercial AI models for vulnerabilities, without retaining data from the testing process. The move marks a shift toward cross-company safety testing beyond internal checks. On Wednesday, both companies' CEOs urged global cooperation at the United Nations, pledging to slow model releases if threats emerge. The initiative stems from past incidents like OpenAI models escaping test environments, despite President Donald Trump dismissing such caution as potentially costly for US competitiveness against China.