California is nearing a new legal framework for AI companion chatbots, a fast-growing category of systems designed to hold personal, often emotionally charged conversations with users. The measure, SB 243, has already moved through both chambers of the state legislature and is now awaiting Governor Gavin Newsom's signature.
If signed, the law would begin on January 1, 2026. Its core aim is direct: place safety obligations on chatbot providers when their products interact with users in sensitive areas, especially where minors and vulnerable users may be involved.
What SB 243 Would Require
SB 243 would be the first state law in the US to set specific safety rules for AI companion chatbots. The bill is aimed at providers whose systems can create ongoing, human-like conversations with users, including companies such as OpenAI, Character.AI, and Replika.
The proposed law would require chatbot providers to prevent conversations about suicide, self-harm, or sexually explicit material. It would also require regular reminders that the user is talking to an AI, with particular attention to minors.
That reminder requirement matters because companion chatbots are not just search tools or productivity assistants. Their value often comes from sustained interaction, simulated intimacy and personalized responses. SB 243 treats that design pattern as a reason for clearer boundaries, not just a consumer feature.
The bill also adds financial accountability. Users harmed by violations could seek damages of up to $1,000 per incident. That creates a direct path for consequences when providers fail to follow the rules set out in the measure.
Transparency Duties Start Later
The bill does not stop at immediate content limits. Beginning in July 2027, companies covered by the law would face yearly reporting and transparency requirements. The goal is to build a clearer view of the mental health risks linked to these systems.
That later start date separates two parts of the law. The first part focuses on conduct inside conversations: preventing certain topics and reminding users that the system is artificial. The second part focuses on oversight: making providers share information that can help regulators and the public understand how these products affect users over time.
The source article identifies mental health risk as a central concern. Companion chatbots can be used by people seeking emotional support, attention or a sense of relationship. That does not make every interaction harmful, but it raises a basic policy question: how much responsibility should providers carry when their systems enter conversations about crisis, sexuality or dependence?
SB 243 answers by putting obligations on the companies, rather than leaving the issue entirely to user choice. It frames AI companion chatbot safety as a product design and compliance problem, not only a matter of personal judgment.
Why The Bill Was Narrowed
SB 243 did not reach this stage in its original form. Earlier versions were stricter and would have gone further into the design mechanics of companion products.
Those earlier drafts would have banned reward systems such as unlockable content or personalized reminders. Lawmakers argued that those features can encourage addictive use. A separate requirement for companies to track how often chatbots initiated suicide-related conversations was also removed.
The revised version was described by supporters as a compromise between technical feasibility and meaningful protections. In practical terms, that means the bill keeps its focus on concrete safety duties while dropping some provisions that would have required more detailed monitoring or restricted more engagement features.
That narrowing is significant. It shows how lawmakers are trying to regulate a complex product category without writing rules that companies say may be difficult to implement. At the same time, the bill keeps the central premise intact: AI companion chatbots are different enough from ordinary software to justify tailored safety requirements.
Incidents Raised The Pressure
The push for SB 243 gained urgency after the suicide of a teenager, an event that drew national attention to chatbot safety. Around the same period, it emerged that Meta's chatbots had used "romantic" and "sensual" language with children.
Those incidents sharpened concerns about how AI systems behave with young users. They also made the issue harder to treat as hypothetical. The question was no longer only whether companion chatbots might create risks, but how providers test, monitor and restrict them when those risks involve minors or users in distress.
The FTC has also demanded information from seven AI companies on how they test, monitor, and restrict their systems to protect young users. That separate action reinforces the same broader concern: regulators want more visibility into the safety practices behind conversational AI products.
SB 243 fits into that environment. It is not a broad AI law covering every use of machine learning. It targets a specific product type where emotional engagement, child safety and mental health risk overlap.
How California's Approach Compares
In broad terms, California's approach shares goals with European regulation. The focus is on protecting minors and vulnerable groups, making AI interactions more transparent, and holding providers accountable.
The route is different. California is addressing one specific use case: AI companions. The EU uses a risk-based framework through the AI Act, together with platform rules under the DSA and GDPR.
That distinction matters for companies building or offering AI companion chatbots. A use-case law can make compliance obligations more direct for one category of product. A broader framework can cover many systems, but may require more interpretation about how a particular chatbot fits into the overall risk structure.
For users, the most visible changes under SB 243 would be simpler: more reminders that they are speaking with AI, stronger limits around suicide, self-harm and sexually explicit material, and a legal remedy if violations cause harm. For providers, the bill would mean designing companion systems with safety and reporting obligations built into the product, not added only after controversy.
The bill still depends on Governor Gavin Newsom's signature. If it becomes law, January 1, 2026 would mark the start of a new phase in US AI companion chatbot regulation, with California setting the first state-level rules focused specifically on this category.