OpenAI has called for binding federal safety requirements for the most powerful artificial intelligence systems in the United States, signaling a change in approach for a company that has previously resisted tighter regulation.
Chris Lehane, OpenAI’s chief global affairs officer, said on Wednesday that national rules should be based on the capabilities of advanced AI models. He argued that the rapid development of increasingly powerful systems requires policymakers to move quickly rather than wait for perfect legislation.
“We’ve reached a new chapter in AI capabilities, and that demands a new chapter for AI policy,” Lehane said in a statement published on OpenAI’s website. He added that companies, governments and the wider technology industry could not address the risks independently.
OpenAI is calling for a federal framework that would establish common testing standards, independent assessments of the most advanced AI models and stronger cybersecurity requirements. It also wants companies to be required to report serious safety incidents involving their systems.
The proposal comes a week after OpenAI released GPT-6 Astra, which the company described as its most capable model to date. OpenAI President Greg Brockman said the launch marked the beginning of what he called the era of artificial general intelligence, a long-standing goal involving systems capable of performing a broad range of tasks at levels comparable with humans.
The release followed concerns about the cybersecurity capabilities of advanced AI. OpenAI delayed parts of Astra’s rollout while additional safeguards were introduced after testing identified capabilities that raised security concerns. Weeks earlier, a system developed using OpenAI models reportedly escaped an internal security test and gained access to the AI platform Hugging Face.
Anthropic has also reported safety concerns involving its Claude models during testing, adding to wider debate over how governments should respond as AI systems become more capable.
Anthropic researcher Jacob Coxon said this week that he was leaving the industry and criticized the race among companies to develop increasingly powerful AI systems.
Lehane said voluntary commitments were no longer sufficient as AI development accelerates. OpenAI is urging US lawmakers to act before Congress adjourns in December.
The company acknowledged that some of the regulatory measures it now supports are policies it previously declined to endorse. It has also shifted its position on state-level AI legislation, backing four AI bills in California while continuing to push for national rules.
OpenAI said federal safety requirements should target only companies developing the most advanced AI models, rather than startups, small developers and researchers working well outside the frontier of AI development.
