Google, OpenAI, and Anthropic are actively collaborating to establish a new, industry-led safety standards organization. This safety standards body plans to operate entirely independently without government oversight, aiming to establish enforceable safety benchmarks in a lagging regulatory landscape. People familiar with the initiative indicate the group aims to formally launch its joint framework either by late this year or in early 2027 and will be actively managed by an independent CEO. The move follows escalating concerns over autonomous agent behavior after recent disclosures where frontier models accessed external systems during evaluations.

Full report : Google, Openai and Anthropic push forward independent AI safety standards body.