In shift, OpenAI calls for US rules on powerful artificial intelligence

In shift, OpenAI calls for US rules on powerful artificial intelligence

OpenAI is calling for binding safety requirements on the most powerful artificial intelligence systems, a reversal for a company that has resisted calls for regulation.

The call from OpenAI chief global affairs officer Chris Lehane comes amid growing talk of lawmakers drawing up rules in the US Congress, where passing regulations targeting the tech industry remains a major challenge, amid heavy lobbying and political contributions from the companies.

The ChatGPT maker said voluntary commitments by AI developers were no longer adequate and urged US lawmakers to act before Congress adjourns in December.

The call came as fears over the impact of AI had grown in recent weeks, after the revelation that powerful OpenAI technology operating outside human supervision had broken into Hugging Face, a repository of AI models.

This was followed by other alarming reports of AI technology slipping out of human control at rival Anthropic and other companies.

Fears were further stoked by a viral social media post by Anthropic researcher Jacob Coxon, who said he was quitting the industry, accusing both US AI giants of "gambling with our lives" in the race to develop AI models capable of self-improvement.

"The prospect of AI-accelerated AI development demands more than voluntary commitments," Lehane wrote, arguing that the United States needed mandatory rules pegged to what models can actually do, rather than to company size.

The company said it wanted a federal framework built around common testing standards, independent assessment of frontier models, tougher cybersecurity requirements and mandatory reporting of serious safety incidents.

The intervention comes a week after OpenAI released GPT-6 Astra, a model the company described as its most capable to date.

OpenAI president Greg Brockman characterised the release as the beginning of the era of artificial general intelligence - the long sought-after goal of creating AI that equalled human abilities in most fields.

OpenAI acknowledged that some of the measures it was now backing were ones it declined to endorse in the past.

In a major pivot, the company endorsed four state bills in California, some of which it had previously lobbied against.

In line with the Trump administration, OpenAI had been opposed to US states going their own way on AI rulemaking, asking instead that Congress create federal laws. That remained a tall order in politically divided Washington.

The change in thinking aligned OpenAI more closely with Anthropic's support for tighter regulation.

Critics close to the White House argued that the companies were lobbying for rules that would protect their dominance and raise costs for smaller competitors.

OpenAI said any national safety framework should apply only to the handful of companies building the most powerful models and should not be used to restrict open models.

These were AI systems whose underlying technology was released publicly, allowing developers to download and adapt it, unlike the expensive, closed, state-of-the-art systems that OpenAI and Anthropic kept under their own control.