OpenAI calls for US-led AI safety standards

OpenAI is urging the United States to organize international safety standards for the most capable AI systems, arguing that national rules alone will not keep pace as machines take on more of the work of improving AI. Its proposal calls for common ways to measure capabilities and safety protections, report serious failures, and coordinate the national institutes that test advanced models. The concern is not today’s chatbots but a possible next step: AI systems helping build their successors, which could speed progress beyond the point where people can meaningfully supervise it. CEO Sam Altman is set to make that case before the UN Security Council; the proposals are not binding rules, but they seek to replace separate company promises with shared, measurable safeguards.