California's AI laws are setting national trend

California's recent wave of artificial intelligence legislation is already shaping the national landscape, with other states moving to adopt similar rules. But they are not copying the Golden State's rules word-for-word--and some are testing the limits of their powers in ways not tried here.

"While other states aren't adopting California's laws wholesale, they are clearly borrowing key principles and replicating California initiatives," said Michael W. M. Manoukian, who represents employers as a partner with Lathrop GPM LLP in San Jose. "Once again, California is acting as a legal bellwether--advancing bold, carefully tailored policies that are shaping the national conversation on AI."

New York, Colorado and Texas are among the states that have passed or proposed laws like ones passed or debated in California. These developments come amid an ongoing lack of federal laws and the defeat of a proposed amendment in the federal One Big Beautiful Bill Act that would have circumvented state-level AI laws.

"States like New York are starting to mirror California's new AI law, with the RAISE Act requiring developers to prepare safety protocols, file incident reports, and disclose risk-mitigation plans similar to California's SB 53," said Aodhan Downey, western region state policy manager at the Computer & Communications Industry Association.

SB 53 was probably the most prominent of several AI laws that Gov. Gavin Newsom signed this year. Dubbed the Transparency in Frontier Artificial Intelligence Act, the law attempts to establish guardrails around the most advanced AI systems, introducing safety protocols and whistleblower protections before they become widely available.

The law represents a compromise. One major AI company, Anthropic PBC, endorsed it. SB 53 incorporated ideas suggested by an AI task force Newsom convened after vetoing a 2024 bill by the same author, Sen. Scott Wiener, D-San Francisco. In a veto message for Wiener's SB 1047, Newsom wrote that the bill imposed overly stringent rules that could curtail the growth of the industry in California.

Newsom also signed SB 243, which puts safeguards on AI chatbots that interact with children, and AB 1043, requiring age verification to use some platforms.

"I expect more states will act following California's lead," Adeel Khan, founder and CEO of MagicSchool, a company that provides AI literacy tools and custom education chatbots, responded in an email. "Illinois, Nevada, and Utah have already limited the use of AI chatbots as mental health substitutes, and California has now set an important precedent with its safety standards for these systems. States that move quickly and thoughtfully have a chance to define what responsible AI looks like and may influence how federal guidelines take shape."

Newsom has repeatedly warned that he does not want to alienate a lucrative industry centered in California. The governor has frequently claimed that 32 of the top 50 AI companies are headquartered in the state.

But attorney Patricia Brum said other states are taking different approaches, some of which might be less onerous to the industry.

"Perhaps in Colorado, they're a little bit ahead of us," said Brum, a partner with Snell & Wilmer LLP in Los Angeles. "I don't know if we are the leading jurisdiction on this issue. For example, in Colorado, the attorney general's office is maintaining a public list of universal opt-out mechanisms."

By contrast, Brum said California tends to put the burden on technology companies. For instance, among the AI-related laws Newsom signed is AB 566, which requires internet browsers to include clear opt-out methods for data collection. These concerns long predate the current worries about generative AI, but the bill's supporters argued that AI has both made it easier for companies to gather people's personal data and made that data more valuable.

Brum said the state's aggressive stance on technology legislation has opened the door to legal challenges that could claim the state has exceeded its jurisdiction.

"You have the Legislature telling California companies--because web browsers are mostly located in California--that they need to implement this, and it technically would be applied to consumers all over, not only in California," Brum said.

Potential litigation may have played a role in Newsom's decisions to veto several bills. These included AB 1064, the Leading Ethical AI Development for Kids Act, which Newsom said was so strict about banning certain kinds of content that it could effectively bar children from using AI tools. It was introduced by Assemblymember Rebecca Bauer-Kahan, D-Orinda, who has been a persistent foe of some in the technology industry and whose bill became a major lobbying target.

Colorado has also passed legislation that will directly affect how companies must behave. The 2024 Colorado AI Act bans discriminatory use of AI, though this year lawmakers delayed implementation until next summer.

New York has looked inward, with lawmakers focusing on transparency in government uses of AI--for instance, in benefits decisions.

Even regulatory-wary Texas has gotten in on the act. In June, Gov. Greg Abbott signed the Texas Responsible Artificial Intelligence Governance Act. Better known as TRAIGA, it created an Artificial Intelligence Council and places significant limits on business and government use of AI. However, the law also centralizes enforcement with the attorney general's office, without the private rights of action included in California's SB 243.

While the approach taken in red states may be different, AI regulation is a bipartisan issue. This summer, the U.S. Senate voted 99-1 to strip the AI Enforcement Pause--a five-year moratorium on most state-level AI laws--out of the One Big Beautiful Bill Act.

While their authors may not advertise it, many of these red-state bills draw on ideas previously proposed in California, Manoukian said.

"Tennessee's ELVIS Act (Ensuring Likeness Voice and Image Security Act), which addresses unauthorized AI-generated voice and image use, parallels California's proposed protections for digital likeness," he said. "Wisconsin and Texas have also enacted election-related AI disclosure laws that echo California's early focus on transparency and deepfake regulation in political communications."