Consensus Yet to Form on California’s Approach to AI Regulation

The state of California tends to lead the way in applying progressive policies to emerging business issues, so should come as no surprise that lawmakers in the Golden State are trying to address some of the landmines around artificial intelligence. Meanwhile, the biggest names in the technology industry are asking for stricter regulation of AI applications. So Big Tech should be supporting legislation intended to do just in California, right? Well, not exactly.

Building support for policy governing potentially lucrative businesses like AI rarely comes easily, and all indications are that this will be no exception.

The California AI bill, SB 1047, primarily seeks to place a higher burden of responsibility on developers for the safety of their AI models. Under the bill’s measures, any developer that spends more than $100 million on an AI model must undertake safety testing and put safeguards in place around the model. Additionally, operating software would need to include a kill switch for AI models. In cases where an AI model causes “severe harm,” the bill grants California’s state attorney general the authority to act against its developer.

SB 1047 is currently awaiting a vote in the California State Assembly after previously passing the state Senate by a nearly unanimous vote. If the legislation clears that hurdle, it will go to Democratic Governor Gavin Newsom for his signature. The sponsor of the bill in the California legislature, Democratic Senator Scott Wiener, has made it clear he wants to get a firm grip on AI while the technology is still in its relative infancy. The thinking among supporters seems to be that the longer we wait to beef up regulation of AI, the tougher it will be to tame.

The European Commission came to a similar conclusion in 2021 when it proposed the first-ever EU regulatory framework for AI, which it claimed would ensure better conditions for the technology’s development and use. The framework, which included a tiered regulatory system based on the risks posed by certain models and applications, was codified into the Artificial Intelligence Act, which was adopted in March 2024.

Although, in theory, Silicon Valley wants strong safety measures for AI, that doesn’t amount to a full-throated endorsement of SB 1047. Google, Meta and OpenAI, for example, have all expressed their reservations about the bill to Wiener. OpenAI said it would prefer a national set of standards for the use of AI, rather than a series of state regulatory regimes. Other prominent players in Big Tech have suggested the bill could have a chilling effect on innovation, with Meta claiming the bill could make California unfavorable to AI development and deployment, possibly causing innovators to flee the state.

For his part, Wiener acceded to some requests to dial back some of the bill’s measures. That made the legislation more palatable to some tech executives, but the bill still has some high-profile detractors on the national political scene. They include former Democratic Speaker of the House Nancy Pelosi, a San Francisco resident, and Democratic Reps. Ro Khanna and Zoe Lofgren from California.

The California bill does have at least one surprising advocate in Elon Musk. The Tesla founder and owner of social media platform X usually takes these kinds of opportunities to troll the establishment. Yet, Musk posted earlier this week that the California bill fits with his longstanding beliefs about the fundamental need for AI regulation. He noted the same standard should apply to “any product/technology that is a potential risk to the public.”

It’s tough to tell if Musk’s endorsement will make SB 1047 more – or less – popular with the public. (You can never know with Musk these days.) Either way, his opinion doesn’t seem likely to derail the AI bill from its path to Newsom’s desk.

***

The Intelligize blog is on hiatus for the Labor Day holiday and will return on Thursday, September 5, 2024