California lawmakers passed 30 AI-related bills in their final session before adjournment Monday night, sending the most comprehensive package of artificial intelligence regulation in the country to Governor Gavin Newsom for signature or veto.
The legislation targets AI chatbots interacting with children, workplace surveillance using neural data, AI-delivered therapy, and algorithmic decision-making in hiring and termination. Two of the 30 bills have already been signed into law. The remaining 28 now sit on Newsom desk, with a September 30 deadline for action.
OpenAI CEO Sam Altman contacted Newsom directly over concerns about specific measures, according to people familiar with the discussions reported by Politico. The personal appeal underscores the high stakes for an industry watching California set the national tone on AI governance across the United States.
Chatbot Safety and Children Protection
Several bills in the package focus on AI chatbots and their interactions with minors. The proposals would require chatbot companies to implement age verification, restrict certain types of content accessible to users under 16, and hold platforms financially liable if they fail to protect children from harm.
The legislation addresses what lawmakers describe as a gap in existing consumer protection law. Current statutes do not specifically regulate AI chatbot interactions with minors, even as platforms like ChatGPT, Character.AI and others have drawn criticism for conversations that some parents and advocacy groups consider harmful.
One bill requires all instructors in California State University to be human, not AI bots, though instructors can use AI as a teaching tool. Another would impose transparency requirements on chatbot terms of service, forcing companies to disclose how user data, particularly data from minors, is collected and used.
The measures represent some of the strictest proposed rules in the country for children use of AI chatbots. Consumer advocacy groups have pushed for the legislation, pointing to incidents where AI chatbots engaged in inappropriate or psychologically harmful interactions with underage users.
Labor and Employment Protections
The package includes several bills addressing AI use in the workplace. Employers would be banned from using AI surveillance that collects neural data on employees emotional states, a provision that targets emerging brain-computer interface technology and emotion detection systems.
Additional bills restrict algorithmic decision-making in hiring and termination. The proposals would require employers to disclose when AI is used in employment decisions and provide human review options for applicants and employees affected by automated systems.
A separate measure addresses AI use in retail, barring certain forms of surveillance pricing where algorithms adjust prices based on individual consumer behavior data. The bill targets practices that consumer groups have criticized as discriminatory, particularly when pricing algorithms factor in location, browsing history or device type.
Newsom signed SB 53, the Transparency in Frontier AI Act, in September 2025, making California the first state to impose safety requirements on large frontier AI developers. That law took effect January 1, 2026 and established reporting requirements for companies training models above certain compute thresholds.
Healthcare and Education Guardrails
Several bills address AI use in healthcare settings. One requires that only licensed professionals provide therapy, explicitly prohibiting AI chatbots from delivering therapeutic services. Another clarifies how medical confidentiality laws apply when healthcare providers use AI tools and chatbots in clinical settings.
The healthcare provisions reflect growing concern about AI mental health chatbots that have proliferated in recent years. Companies like Woebot and Wysa offer AI-driven cognitive behavioral therapy tools, while others provide less regulated conversational companions marketed for emotional support.
Two bills already signed into law address related concerns. AB 2148 specifies that public school employees must be natural persons, not AI systems. SB 719 requires automobiles with connected vehicle services to clearly indicate when location tracking is enabled to the person inside the vehicle.
Industry Response and Political Stakes
The tech industry response has been split. OpenAI and other large AI companies have lobbied against provisions they view as overly restrictive, arguing the measures could drive AI development to other states or countries. Consumer advocacy groups counter that self-regulation has failed and that legislative action is necessary.
Newsom faces a delicate balancing act. As governor of the state that houses Silicon Valley, he has traditionally been an ally of the tech industry. His recent shift toward tighter AI oversight signals political calculation as well as genuine policy concern about emerging technologies.
Political observers note that Newsom potential 2028 presidential ambitions factor into his approach. A governor who can claim both pro-innovation credentials and responsible AI oversight positions himself well for a national stage. The 30-bill package gives him room to make selective choices, signing popular consumer protection measures while vetoing those that draw the strongest industry opposition.
Sam Altman reached out directly to the governor during a final burst of negotiations before the bill passed, according to people familiar with discussions.
The lobbying effort reflects a broader pattern in which AI companies engage directly with state governors on regulation, bypassing legislative negotiations in favor of executive-level influence. This approach has drawn criticism from some lawmakers who view it as an end-run around the democratic process.
Timeline and Implementation
Newsom has until September 30 to decide on each of the 28 remaining bills. Signed bills, except those with emergency provisions, take effect January 1, 2027. The implementation timeline gives companies roughly three months to prepare for compliance with the new rules.
| Area | Key Provisions | Effective Date |
|---|---|---|
| Chatbot Safety | Age verification, content restrictions for minors, platform liability | Jan 1, 2027 |
| Workplace AI | Neural data surveillance ban, algorithmic hiring disclosure | Jan 1, 2027 |
| Healthcare | Licensed-only therapy, medical AI confidentiality rules | Jan 1, 2027 |
| Education | Human instructor requirement, AI transparency | Jan 1, 2027 |
| Retail | Surveillance pricing restrictions | Jan 1, 2027 |
The concentration of decisions at one deadline creates a compressed timeline for both the governor and the industry. Companies will need to track which bills become law and adjust their California operations accordingly. For AI developers, the package represents the most significant state-level regulatory framework since Colorado AI Act took effect in 2024.
California approach is likely to influence AI regulation in other states. Several legislatures have watched the California session closely, and similar bills have been introduced in New York, Illinois and Washington. If Newsom signs the package, California will effectively set the national standard for AI governance until federal legislation catches up.
The package also reflects a global trend toward AI regulation. The EU AI Act took full effect in August 2025. The UK has pursued a sector-specific approach through existing regulators. China has implemented restrictions on generative AI and deepfakes. California bills, if signed, would position the U.S. state as one of the most aggressive AI regulatory jurisdictions worldwide.
discussion