OpenAI backs measure that would require independent audits of AI models

ChatGPT maker OpenAI supports parts of a bipartisan bill in Congress that would require the largest AI companies to submit their large language models to independent audits, according to Chris Lehane, the company’s chief global affairs officer.

OpenAI has previously backed state legislation to regulate AI, but this marks the first time the company has thrown its weight behind a federal mandate that would require a third-party group to assess AI developers’ work. 

Lehane expressed support for the proposal, which is included in a bill now before lawmakers called the FRONTIER Act, during a roundtable with lawmakers in Washington, D.C., on Tuesday, an OpenAI spokesperson told CBS News. 

Specifically, an independent organization would have to determine that an AI company’s safety protocols are sufficiently rigorous to mitigate what the bill describes as “acceptable levels of catastrophic risk.” 

Lehane’s comments come as AI researchers and developers warn about the potential for advanced AI to inflict harm on people. 

The FRONTIER Act, sponsored by Reps. Jay Obernolte, a Republican from California, and Lori Trahan, a Democrat from Massachusetts, would also require AI developers to publish reports on their models and disclose safety incidents. 

Although OpenAI supports third-party audits of AI models, the company hasn’t endorsed the bill in its entirety, the spokesperson noted. 

In a Sept. 9 blog post, Lehane called for regulatory action to control AI development. OpenAI is calling for Congress to enact mandatory national AI safety requirements and advocates that leading AI developers collaborate to develop industry standards. 

“The stakes are enormous. Advanced AI could accelerate the development of new medicines, strengthen critical infrastructure, expand economic opportunity, and help solve scientific problems that have resisted generations of human effort,” Lehane wrote. “But the capabilities that make models more useful also come with risks, and they will not remain confined to a few frontier laboratories.”

Joseph Miller, an AI safety researcher and founder of PauseAI UK, a group that supports slowing AI development, said that the public debate over how to control potential AI risks is long overdue. He also thinks leading AI developers are unlikely to take sufficiently strong measures to safeguard their technology absent government intervention. 

“We need government to lead on it, because they probably aren’t going to do it without the government stepping in,” Miller told CBS News.