California's AI Bills Face Final Judgment
Key highlights this week:
We’re tracking 1,093 bills in all 50 states related to AI during the 2025 legislative session.
Lawmakers failed to come to an agreement to amend Colorado’s major AI law during a special session, and instead delayed the law’s effective date by a few months to June 2026.
Attorneys general in over 40 states warn of potential harms from deepfakes and companion chatbots.
Industry is forming super PACs to combat AI regulations in states like California, as over a dozen AI-related bills in the Golden State could be headed to the governor's desk this month, which is the topic of this week’s deep dive.
As California’s legislative session winds down, attention is once again on the Golden State to see whether it will maintain its leadership role in regulating emerging tech issues like artificial intelligence. Dozens of proposals were axed last week, but many key measures remain under consideration as lawmakers enter the final stretch. The surviving AI bills cover a wide range of priorities, from regulating foundation models and algorithmic pricing to protecting children from harmful chatbot interactions.
This year, lawmakers in California introduced 47 AI-related bills. (Surprisingly, Montana (48), Illinois (48), New Jersey (67), Texas (81), Massachusetts (93), and New York (143) all introduced more AI-related bills this session.) But with just a few weeks to go, only 17of California’s AI bills remain alive, with several held up in the “suspense file” on Friday. Below, we run through the bills that have survived to this point and are still eligible for lawmakers to send to Governor Newsom (D) before the legislature adjourns on September 12.
Algorithmic Bias and Automated Decision Systems
Following the lead of Colorado, lawmakers spearheaded a big push earlier this year to pass broad legislation to mitigate bias when AI is used in “consequential” decisions in “high-risk” areas impacting consumers. However, these efforts failed to enact a major algorithmic discrimination law this year. In Virginia, lawmakers passed a version that Gov. Youngkin (R) vetoed. A proposal in Texas was scaled back, and efforts fell short in Connecticut once again. Even Colorado’s landmark algorithmic discrimination law was further delayed during last month’s special session, with a new effective date of June 2026 and likely amendments scaling back its requirements before it ever goes into effect.
But an algorithmic discrimination bill is still making its way through the legislative process in California. Assemblymember Rebecca Bauer-Kahan’s bill (CA AB 1018) would regulate automated decision systems used to facilitate “consequential decisions,” such as those involving employment, housing, education, and health care. The measure has been heavily criticized by industry groups and the California Chamber of Commerce for being overly broad and vague.
AI Model Safety
Last year, Sen. Scott Wiener’s (D) SB 1047, which focused on the safety of AI models themselves, gained a ton of headlines but was ultimately vetoed by Gov. Newsom. This year, Sen. Wiener returns with a new AI safety bill (CA SB 53) that would regulate foundational models. Wiener incorporated some of the findings from Governor Newsom’s AI working group and has found less intense industry pushback than last year’s bill (so far). A similar bill (NY AB 6453 / SB 6953), the RAISE Act in New York, is sitting on the governor’s desk. But Gov. Hochul (D) can wait until the end of the year to make a final decision on whether to sign the bill or not, and she’ll likely wait to see how Gov. Newsom acts before making a decision.
Additionally, Assemblymember Maggy Krell’s (D) bill (CA AB 316) would prohibit a developer from asserting a defense in a civil case that artificial intelligence autonomously caused the harm to the plaintiff. The bill was amended last week to clarify that defendants can still present other affirmative defenses, or defenses relevant to comparative fault.
Employer AI Use
A Senate bill known as the “No Robo Bosses Act” (CA SB 7) would regulate the use of automated decision systems by employers, requiring written notice to employees. Major labor unions in the state have pushed for the bill, but it has been watered down in amendments to exclude hiring decisions from general requirements, narrow the restriction on employer uses, and narrow the scope of employee appeals.
Children’s Safety with AI
High-profile lawsuits against companion chatbot developers for the suicides of some users have led to increased scrutiny by lawmakers. The Assembly has already passed a measure known as the Leading Ethical AI Development (LEAD) for Kids Act (CA AB 1064) that would prohibit developers from producing a companion chatbot intended to be used by or on a child. The Senate passed a similar bill (CA SB 243) that requires companion chatbot platforms to take reasonable steps to prevent prolonged engagement and to implement protocols for addressing thoughts of suicide or self-harm.
Algorithmic Pricing
Retailers and landlords have increasingly turned to algorithmic tools to set prices, raising concerns that these systems can collude to keep costs artificially high or use invasive data to charge different customers different rates. Two bills (CA AB 325 and CA SB 295) would prohibit using or distributing a pricing algorithm that uses nonpublic competitor data. Another bill (CA SB 259) would prohibit what’s known as “surveillance pricing,” where personal or behavioral data is used to adjust what a consumer pays. Another bill (CA AB 446) was amended last week to ban surveillance pricing, but only in grocery stores, and only prohibiting price increases, not discounts.
Health Care
In addition to companion chatbots, some users have turned to generative AI for online therapy. A proposed bill (CA AB 489) would prohibit generative AI from implying that any care or advice offered is provided by a licensed person, subject to sanctions from the appropriate health care profession board.
Fears over discrimination by AI models prompted a bill (CA SB 503) that requires developers and deployers to make reasonable efforts to mitigate the risks of reasonably foreseeable biased impacts in health care decision-making. Beginning in 2030, developers would be required to submit systems to an independent third-party auditor.
Deepfakes
The ability of AI to produce the likeness of users and famous figures has raised concerns about how that can be used for nefarious purposes. A proposal (CA SB 11) would require AI technology that enables a user to create a digital replica, to provide a consumer warning that non-consensual production of digital replicas could lead to civil or criminal liability. The bill also clarifies that the use of a digital replica with the intent to impersonate another is a false personation.
California’s political deepfake law was recently struck down by a federal judge, so a bill amending that law (CA AB 502) seems unlikely to go anywhere this session. The bill proposed this year would have regulated the disclosure text for a political deepfake and authorized a depicted individual to bring a civil action.
State Use
Also up for consideration are bills that would regulate how the government uses AI. One bill (CA SB 524) would require law enforcement officials to disclose when AI is used to write reports with required documentation. The state budget bill (CA SB 102) would require agencies that use generative AI to complete risk assessments.
Bills That Didn’t Make It
However, many AI bills failed to pass out of committee on Friday. These AI bills that are dead for the year, but could be brought back next year, include:
CA AB 410, which would have required disclosure of chatbot interactions with consumers
CA AB 412, which would have required disclosure of copyrighted materials used to train AI
CA AB 621, which would have created liability for a service that allows deepfake sexual content
CA AB 853, which would have required online platforms, devices, and generative AI systems to label synthetic content
CA AB 1405, which would have regulated AI auditors
CA SB 52, which would have prohibited rental algorithms
CA SB 69, which would have directed a program to develop state expertise in AI
CA SB 384, which would have prohibited price-setting algorithms
CA SB 420, which would have regulated developers and deployers of high-risk automated decision systems and prohibited discrimination
CA SB 833, which would have regulated AI used for critical infrastructure
What Will the Governor Do?
As lawmakers race toward adjournment, it remains unclear whether Gov. Newsom will ultimately support any of the surviving measures, especially given his caution on last year’s veto. Even if some bills do pass, their fate could be complicated by looming federal preemption, leaving California’s role in setting the rules for artificial intelligence uncertain for now.
Recent Developments
Industry News
Super PACs: Last month, the tech industry announced two separate efforts to use political action committees focused on supporting political candidates who will vote against stringent regulations of AI. Meta is backing a super PAC in California with tens of millions of dollars. Separately, AI inventors and an OpenAI co-founder are pouring up to $200 million into “Leading the Future,” a group of super PACs that will back congressional candidates as well as state officials in California, New York, Illinois, and Ohio.
Major Policy Action
National: A bipartisan coalition of 44 state attorneys general sent a letter to AI developers warning of the dangers companion chatbots may pose to children. Another bipartisan coalition of 47 attorneys general called on search engines and payment platforms to take action to mitigate the spread of sexual deepfake content online. The letter urges search engines to prevent searches to find how to create or disseminate sexual deepfake content, and for payment platforms to deny services to providers that allow the creation of sexual deepfakes.
Colorado: During last week’s special session, lawmakers were unable to reach an agreement on changes to the state’s artificial intelligence law, and instead approved a bill (CO SB 4) to push back the effective date from February to June of 2026. Bill sponsor Sen. Robert Rodriguez (D) tried to replace the existing law with a new law that would require disclosures to consumers, but ultimately scrapped those provisions in favor of delaying the effective date until a fix can be found next session.
Massachusetts: Next Thursday, the Joint Committee on Advanced Information Technology, the Internet and Cybersecurity will hold a hearing on a number of artificial intelligence-related bills, including legislation on transparency in AI systems, digital replicas, political and sexual deepfakes, AI use in health care, and energy use due to AI.
North Carolina: Gov. Josh Stein (D) signed an executive order on Tuesday, creating the North Carolina AI Leadership Council to advise and support state agencies on AI strategy, policy, and training. The Department of Information Technology will also house a new “AI accelerator,” which will “explore and implement AI solutions to real world problems,” according to Secretary Teena Piccione.
Virginia: On Wednesday, the Joint Commission on Technology and Science held a hearing to discuss potential artificial intelligence legislation for the 2026 session. Among the topics discussed were labeling requirements for AI-generated content and algorithms for rental properties.
Notable Proposals
Wisconsin: Senate Republicans introduced a bill (WI SB 400) that would prohibit state employees from downloading AI apps owned or operated by an entity located in a foreign country of concern, such as DeepSeek. Alabama, Arkansas, Iowa, Kansas, Nebraska, New York, North Dakota, Oklahoma, South Dakota, Tennessee, Texas, and Virginia are among the states that have already banned such apps from state devices.
