California has moved ahead with a broad set of AI laws after Governor Gavin Newsom considered 38 AI-related bills in September. The most disputed measure, SB 1047, was vetoed on Sunday, but 18 other AI bills have been signed into law.
The result is not one sweeping framework. It is a collection of targeted rules aimed at specific uses of generative AI, including deepfake nudes, election content, automated healthcare decisions, AI-generated robocalls, and digital replicas of performers.
What California chose to regulate first
The new laws focus on areas where AI systems can directly affect people, institutions, and public trust. Some rules require disclosure. Others create reporting systems, limit automation, or expand existing protections to cover AI-generated material.
One law, SB 896, requires California's Office of Emergency Services to perform risk analyses on possible threats from generative AI. CalOES will work with frontier model companies, including OpenAI and Anthropic, to study risks to critical state infrastructure and threats that could lead to mass casualty events.
Another law, AB 2013, turns attention to training data. It goes into effect in 2026 and requires generative AI providers to publish documentation on their websites describing the data used to train their AI systems. That documentation must include dataset sources, how the data is used, the number of data points, whether copyrighted or licensed data is included, the time period when the data was collected, and other standards.
California also clarified that its existing privacy laws apply to generative AI systems. Under AB 1008, if an AI system such as ChatGPT exposes personal information, including a name, address, or biometric data, California's privacy laws limit how businesses can use and profit from that data.
Schools, healthcare, and a legal definition of AI
The education laws are aimed less at punishment and more at preparation. AB 2876 requires California's State Board of Education to consider AI literacy in math, science, and history curriculum frameworks and instructional materials. The source describes this as a path toward teaching students how artificial intelligence works, along with its limitations, impacts, and ethical considerations.
SB 1288 adds another education requirement by directing California superintendents to create working groups that explore how AI is being used in public school education.
California also adopted a uniform definition of artificial intelligence through AB 2885. The definition covers an engineered or machine-based system that varies in autonomy and can infer from input how to generate outputs that influence physical or virtual environments.
Healthcare is another major focus. AB 3030 requires healthcare providers to disclose when they use generative AI to communicate with a patient, specifically when the messages include the patient's clinical information. SB 1120 limits how health care service providers and health insurers can automate services and ensures licensed physicians supervise the use of AI tools in those settings.
Deepfakes, robocalls, and online platforms
Several of the laws target synthetic media that can mislead, exploit, or pressure people. AB 2905 requires robocalls to disclose when they use AI-generated voices. The source links this measure to concern over a deepfake robocall resembling Joe Biden's voice that confused many New Hampshire voters earlier this year.
California also expanded and added rules around deepfake pornography. AB 1831 expands existing child pornography laws to include matter generated by AI systems. SB 926 makes it illegal to blackmail someone with AI-generated nude images that resemble them.
SB 981 places duties on social media platforms. Platforms must establish channels for users to report deepfake nudes that resemble them. The content must be temporarily blocked while the platform investigates and permanently removed if confirmed.
Another law, SB 942, addresses AI-generated content labeling through provenance data. Widely used generative AI systems must disclose that content is AI-generated in that data. The source gives the example that images created by OpenAI's Dall-E need a tag in their metadata saying they are AI generated.
Election content and political ads
California's election-related AI laws focus on deepfakes that could influence voters. AB 2655 requires large online platforms such as Facebook and X to remove or label AI deepfakes related to elections and create channels to report that content. Candidates and elected officials can seek injunctive relief if a large online platform does not comply.
AB 2839 applies to social media users who post or repost AI deepfakes that could deceive voters about upcoming elections. The law went into effect immediately on Tuesday, and Newsom suggested Elon Musk may be at risk of violating it.
AB 2355 requires disclosures for AI-generated political advertisements. The source notes that this means Trump may not be able to get away with posting AI deepfakes of Taylor Swift endorsing him on Truth Social, while also noting that she endorsed Kamala Harris.
Actors, digital replicas, and the SB 1047 veto
Two laws address AI in California's media industry and were pushed by SAG-AFTRA, the nation's largest film and broadcast actors union. AB 2602 requires studios to obtain permission from an actor before creating an AI-generated replica of their voice or likeness.
AB 1836 protects deceased performers. It prohibits studios from creating digital replicas of deceased performers without consent from their estates. The source cites legally cleared replicas used in the recent Alien and Star Wars movies, as well as other films.
The major bill Newsom did not sign was SB 1047. In a letter explaining the veto, Newsom said the bill focused too narrowly on large AI systems that could give the public a false sense of security. He said small AI models could be just as dangerous as the systems targeted by SB 1047 and that a more flexible regulatory approach is needed.
Newsom had also discussed the issue during a chat with Salesforce CEO Marc Benioff at the 2024 Dreamforce conference. His comments framed the regulatory challenge as a question of which AI risks are demonstrable, which are hypothetical, and what can realistically be addressed across the spectrum.