OpenAI backs measure that would require independent audits of AI models

OpenAI, the creator of ChatGPT, has indicated support for certain provisions within a bipartisan bill currently before Congress. According to Chris Lehane, OpenAI's chief global affairs officer, the company has previously backed state-level legislation aimed at regulating AI. However, this marks the first instance where OpenAI has publicly endorsed a federal measure that would mandate independent audits of AI models by third-party organizations.

This support for the bill, which is part of the FRONTIER Act, was expressed during a roundtable discussion with lawmakers in Washington, D.C. The audits would ensure that an AI company's safety protocols are rigorous enough to mitigate what the bill refers to as "acceptable levels of catastrophic risk." Lehane's endorsement comes amid growing concerns among AI researchers and developers about the potential for advanced AI to cause harm to individuals.

While OpenAI supports the concept of third-party audits, it has not endorsed the bill in its entirety. In a September 9 blog post, Lehane called for regulatory action to govern AI development. He and OpenAI are advocating for mandatory national AI safety requirements and collaboration among leading AI developers to establish industry standards.

The potential benefits of advanced AI are significant, ranging from advancing medical research to strengthening infrastructure and economic opportunities. However, the risks associated with these capabilities cannot be ignored, as they may not remain confined to select labs, according to Lehane.

Joseph Miller, an AI safety researcher and founder of PauseAI UK, believes that the public debate over how to manage AI risks is overdue. He believes that while leading AI developers may not take sufficient measures to safeguard their technology without government intervention, government intervention is necessary. Miller's comments were made to CBS News.