OpenAI shifts stance, calls for U.S. legislation to regulate powerful AI

OpenAI is calling on the U.S. to impose binding safety requirements on the most powerful AI systems and urging Congress to act before its December recess, saying developers’ voluntary commitments are no longer enough. The company argues that a federal framework should include common testing standards, independent evaluations of frontier models, stricter cybersecurity requirements, and mandatory reporting of serious safety incidents. The previous week, OpenAI released GPT-6 Astra and acknowledged that some of the measures it now supports are ones it had previously refused to endorse.