California Writes the Playbook for AI Regulation
Key Takeaways
- California is shaping up to be the most active state on AI regulation in 2025, with new laws, pending bills, and agency regulations all moving forward at the same time.
- Gov. Newsom signed CA SB 53 into law, making California the first state to require developers of large frontier AI models to proactively address safety concerns through a published framework and annual updates.
- The California Privacy Protection Agency finalized rules on automated decision-making technology, giving consumers the right to notice and opt-out when AI is used in high-stakes decisions like hiring, housing, and healthcare.
- The "No Robo Bosses Act" (CA SB 7) is still awaiting the governor's signature and would require employers to notify workers when AI influences employment decisions and mandate human review before any AI-driven termination.
- The broader legislative landscape has shifted: the AI safety approach is gaining ground in New York and California, while the algorithmic discrimination model pioneered by Colorado has stalled, with SB 205 likely facing major changes before it even takes effect.
- If you're a subscriber, click here for the full edition of this update. Or, click here to learn more about our MultiState.ai+ subscription.
As we enter the final months of 2025, California is poised to make the biggest changes regarding AI regulation this year. While notable proposals failed to make it across the finish line in other states, in the last two weeks, California has issued new AI regulations, signed a major AI safety bill into law, and Governor Gavin Newsom (D) still has a substantial AI employment bill awaiting his decision. This comes as two agencies in California have also released final regulations this year addressing AI.
All eyes remain on California as hundreds of bills sit on Gov. Newsom's desk, including 10 AI-related bills. While the governor has until October 12 to make a decision, he signed a major piece of AI legislation into law well before that deadline.
California Enacts Major AI Safety Legislation with SB 53
Earlier this week, Gov. Newsom signed Sen. Scott Wiener's (D) CA SB 53 into law. We previously discussed this bill as it navigated the legislative process this year.
Requirements for Frontier AI Model Developers
The new law will require the developers of large, "frontier models" of AI to develop, implement, comply with, publish an AI framework, and provide annual updates. This is the first law that requires AI developers to proactively address safety concerns in the largest AI models. When signing the bill, the governor noted that it struck a balance between the need to protect communities while allowing the AI industry to safely grow and thrive.
Shifting Regulatory Landscape Across States
A similar bill, the RAISE Act (NY AB 6453/SB 6953), is currently awaiting the signature of New York Governor Kathy Hochul (D), who has until the end of the year to make a decision. If enacted, it may signal a change in how lawmakers believe AI will be regulated. Late last year, it appeared the legislation broadly targeting algorithmic discrimination and consumer protection in "high-risk" decisions was on the verge of spreading after Colorado enacted SB 205. At the same time, Gov. Newsom had recently vetoed Sen. Wiener's SB 1047, setting back the proponents in the AI safety camp. But those fates have reversed this year, with this year's version of the AI safety bill becoming law in California and no state following Colorado's lead on SB 205. In fact, SB 205 itself was delayed and will likely be substantially amended next session before even going into effect. At least that's the case legislatively.
New Automated Decision-Making Technology Regulations Take Effect
Shortly before SB 53 was signed into law, the California Privacy Protection Agency (CPPA) issued final rules for automated decision-making technology. The regulations, issued under the California Consumer Privacy Protection Act, require businesses to give notice when using automated decision-making technology to make significant decisions and to provide consumers the ability to view and opt out of the data collection used by these technologies. Similar to many of the high-risk AI bills considered in many states this year, the regulations define a "significant decision" as one "that results in the provision or denial of financial or lending services, housing, education enrollment or opportunities, employment or independent contracting opportunities or compensation, or healthcare services."
Notably, these regulations impact the type of decision-making technology at issue in many high-risk AI bills. However, instead of requiring reports and audits of these technologies, the regulations seek to protect consumers by effectively allowing them to opt out of having AI used to make decisions that can impact their daily lives.
Pending Employment AI Legislation and Workplace Protections
One of the more notable AI bills still sitting on Gov. Newsom's desk is CA SB 7, also known as the "No Robo Bosses Act," which regulates AI's use in employment decisions.
Key Provisions of the No Robo Bosses Act
The bill requires employers to provide written notice when AI is used to make employment decisions that materially impact an individual, such as hiring, compensation, responsibilities, scheduling, and discipline, including termination. Additionally, the bill prohibits relying solely on AI to make termination decisions. When AI is relied on primarily to make discipline or termination decisions, that decision must be reviewed by a human.
This bill applies to traditional employees as well as independent contractors, and the language prohibiting the use of AI as the sole source of termination decisions and the requirement to have those decisions reviewed by a human also applies to deactivation decisions. The inclusion of independent contractors is notable, given the state's past fights over worker classification, and ensures that this bill, if signed into law, would impact most California employers.
California Civil Rights Council Workplace AI Regulations
This bill also comes before the governor as regulations addressing AI in the workplace, previously approved by the California Civil Rights Council, take effect on October 1. The regulations aim to protect against employment discrimination as a result of the use of AI, prohibit using AI to discriminate against applicants or employees based on a class protected under the California Fair Employment and Housing Act, and provide clarity on how existing anti-discrimination laws apply to AI used in employment decisions.
Regardless of what further action Gov. Newsom takes on the remaining AI-related bills on his desk in the coming weeks, California has taken substantial steps in regulating AI this year, and we'll need to wait until January to see if other states will take a similar approach.
If you're a subscriber, click here for the full edition of this update. Or, click here to learn more about our MultiState.ai+ subscription.
