After making a major push to further regulate Artificial Intelligence (AI) last year, the California Legislature seems content to nibble around the edges as session comes to a close. The latest measures mostly add unnecessary regulations and increase liability for tech and business firms, epitomizing a “death by a 1,000 cuts” approach rather than a far-reaching regulatory strategy.
One bill, Senate Bill 1000, would require AI developers to disclose if generative AI created or altered content. Another, Senate Bill 300, requires “chatbot platforms to establish protocols to detect, remove and respond to instances of suicidal ideation, suicide or self-harm expressed by users.” The latter is particularly problematic as it would impose a strict liability standard that — and what else is new? — could lead to an avalanche of lawsuits.
Assembly Bill 2545 would mainly expand the state’s bureaucracy by creating a 14-member advisory board within the Employment Development Department to review “existing data collection systems and gaps in data collection related to the use and impact of advanced AI systems.” That’s the same EDD that handed out billions of dollars in improper COVID-related payments, so we’re skeptical it can successfully oversee the complexities of AI data gathering.
Assembly Bill 2575 is a labor-oriented measure that limits healthcare facilities’ use of AI that, per the committee analysis, is designed to address “concerns from health care workers that deployment” of AI systems “may override their professional judgment and that their decision to exercise their own judgment … might subject them to adverse treatment in the workplace.”
In August, the first part of the “California AI Transparency Act” goes into effect. It requires large AI firms to label their photos, videos and images to protect people from thinking they might be real. California isn’t the only state that’s releasing AI proposals and laws in dribs and drabs, as legislators nationwide have introduced an estimated 1,800 AI-related bills this year.
Last fall, the industry dodged a bullet as one of the highest-profile measures, Assembly Bill 1018, stalled on its third attempt. That bill epitomizes the state’s dated approach toward AI regulation. The Legislature wants to recycle the heavy-handed regulatory approach embraced by the Biden administration and the European Union.
The bill targeted Advanced Decision Systems. Per one committee analysis, “relying on ADS to make consequential decisions can be hazardous if the systems are not trained carefully or tested thoroughly.” Supporters — mostly unions and progressive activists — fear AI systems will pre-empt humans on “apartment leases, school admissions, and, in the workplace, hiring, firing, promotions and disciplinary actions,” CalMatters explains.
According to opposition from a business-community coalition, the bill exposed “smaller businesses to significant — if not devastating — liability even for mere errors that caused no harm to consumers, but it would also hinder many beneficial uses of ADS.” Unfortunately, we can expect that measure to come back next session.
We expressed optimism at the Trump administration pro-growth approach to AI, but it has recently taken a detour toward a regulation-oriented federal approach in service to its national security and protectionist priorities. That’s too bad, as California’s commitment to the old Biden strategy speaks to the need for a sensible set of innovation-focused national standards.
We’re seeing some bipartisan efforts in Congress, but it’s not yet clear if enough lawmakers are listening.