STEVE INSKEEP, HOST:
California could soon impose safety standards for artificial intelligence. These would be the first in the nation. Lawmakers have sent Governor Gavin Newsom a bill that would require the makers of the most powerful AI models to test them for potential harm. The tech industry is divided and so are politicians. From KQED in San Francisco, Rachael Myrow reports.
RACHAEL MYROW, BYLINE: It's one of dozens of AI-related measures introduced in California this legislative session, including efforts to regulate political deepfakes, tackle bias in decision-making and protect children's data. But Democratic State Senator Scott Wiener of San Francisco has arguably introduced the most contentious AI bill of the session.
SCOTT WIENER: Doing it this way does allow us to take a more methodical approach in terms of not having to solve everything all at once.
MYROW: Wiener's bill would require developers to ensure their programs don't lead to impacts that AI leaders have publicly warned against, like cyberattacks on critical infrastructure or turbocharging the development of biological weapons. The measure would only affect companies developing AI systems that cost more than a hundred million dollars to train. If their tech is used to harm people, they could be sued by the California attorney general. Wiener has spent the last year in a pitched battle over the bill with some of the most prominent figures in generative AI.
ANDREW NG: This would drive a lot of lawyers' fees but ironically without making AI safer.
MYROW: Stanford professor and former Google executive Andrew Ng warns the policy could strangle the nascent industry in its cradle.
NG: The crux of the issue is this tries to regulate a general-purpose technology in a way that is very difficult or impossible for the providers of that technology to really know what exactly they're supposed to do.
MYROW: Democratic House Speaker Emerita Nancy Pelosi is one of several Silicon Valley representatives in Congress who have come out against the bill. She told KQED the measure is well intentioned but ill informed.
NANCY PELOSI: We protect what California does. But I don't want California going down a bad path on something this serious.
MYROW: That said, there's not consensus that Wiener's measure is a bad path in AI circles or in politics. Even Elon Musk has defended the bill. On Monday, he posted on X that AI should be regulated like any other product or technology that's a, quote, "potential risk to the public." What's next? Governor Gavin Newsom's office has not indicated which way he's leaning. Here's bill sponsor Wiener again.
WIENER: My experience with Governor Newsom is he gives bills a fair shake, and he speaks to people who support and people who oppose, and he makes an informed choice. And I'm confident he will do that here.
MYROW: The governor has until September 30 to sign or veto the bill.
For NPR News, I'm Rachael Myrow in San Francisco. Transcript provided by NPR, Copyright NPR.
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.