From Chatbots To Police Reports: 6 New California AI Laws To Know

Gov. Gavin Newsom signed an executive order Thursday directing the state to prepare workers, businesses, and communities for economic disruption brought about by artificial intelligence. The order directs state agencies, experts, academics, and industry leaders to gather data, develop new policies, and identify early warning signs of workforce disruption.

The order directs the state to explore policies including severance standards, employment insurance and transition support for displaced workers, worker ownership models, universal basic capital concepts, expanded workforce training, and stronger tracking of hiring and payroll trends to help the state respond faster to potential layoffs and economic disruption.

Subscribe

California has already passed dozens of laws and regulations related to artificial intelligence, which range from chatbot disclosures to deepfake prohibitions to regulations on the use of AI in hiring and employment decisions. Supporters, including child safety advocates, labor groups, and consumer privacy organizations argue the laws provide necessary guardrails as AI expands into every aspect of contemporary life. Tech companies and industry groups have argued that some rules could stifle innovation, create compliance burdens, or force the release of proprietary information.

Here are six important AI laws California enacted or taking effect over the past year:

1. SB 53: Transparency in Frontier Artificial Intelligence Act

This landmark law, which went into effect in January, requires developers to publish redacted risk evaluations and safety protocols for large AI models. Under SB 53, tech companies could face penalties of up to $1 million if they do not report critical safety and security incidents – including major cyberattacks or loss of model control – to the California attorney general within 15 days. SB53 also establishes mechanisms and legal protections, including payment of attorney’s fees, for employees and contractors to anonymously report catastrophic risks or false statements from their employers.

2. AB 853: California AI Transparency Act

AB 853, signed into law in October 2025, requires large AI companies and online platforms, including social media companies, to implement tools and disclosures to help users identify AI-generated or altered images, audio, and video. Starting in August 2026, large AI providers must offer AI detection tools allowing users to assess whether or not content is AI-generated or altered. In January 2027, these rules will also apply to online platforms and social media companies. By January 2028, cameras, phones, and recording devices will be required to offer tools to help users prove they were not AI-generated.

3. SB 243: Companion chatbots

Passed in October, SB 243 is the first comprehensive state law to require companies to disclose to users that they are interacting with an AI companion chatbot (the law does not apply to customer service chatbots). When interacting with users known to be minors, operators must provide a clear and conspicuous notification at least every three hours during ongoing interactions. All platforms must also disclose that companion chatbots may not be suitable for minors. Operators must also maintain protocols to prevent suicide or self-harm, including referring users to crisis service providers when necessary. Starting in 2027, operators must annually report statistics and protocols to the California Office of Suicide Prevention.

A stricter proposal known as the LEAD for Kids Act would have imposed broader restrictions on AI companion chatbots used by minors, especially those that could encourage self-harm, engage in sexual conversations, promote dangerous behavior, and other harmful behaviors. Gov. Newsom vetoed it, arguing it was too broad and could effectively block many chatbot products for minors.

4. AB 2013: Generative Artificial Intelligence Training Data Transparency

AB 2013, which went into effect in January, requires generative AI companies to provide public summaries detailing how they train their systems. The public summaries must disclose where training data comes from, whether copyrighted material was used, whether personal information was involved, whether datasets were purchased or licensed, whether synthetic data was used, and more.

This law has proved controversial with AI companies, who argue that training data is proprietary information. Elon Musk’s xAI sued for that reason, though a judge refused to block the law from taking effect.

5. AB 621: Deepfake pornography

AB 621, effective October 2025, prohibits “deepfake pornography,” defined as digitally altered, sexually explicit material depicting individuals without their consent. Anyone who knowingly creates, shares, or facilitates this content can face damages of $1,500 to $50,000 per violation, and up to $250,000 for malicious conduct. The law offers limited exemptions for protected speech, journalism, and legal proceedings.

Another law, SB 857, also criminalizes the knowing possession, distribution or production of obscene or sexualized images using AI or digital alteration, even if no real child was involved.

6. SB 524: AI in Law Enforcement Oversight

SB 524 requires law enforcement agencies to disclose when official reports are generated using artificial intelligence, even partially. Each page of those reports must disclose that AI was used. Records must also show which AI tools were used, the changes that were made, who made them, and what evidence the AI used to generate the report.

SEE ALSO:

• CA Becomes First In Nation To Crackdown On AI Chatbots
• California Advances Measures Targeting AI Discrimination And Deepfakes