California SB 243 (Companion Chatbot Safety Act)
Last reviewed
Jun 7, 2026
Sources
16 citations
Review status
Source-backed
Revision
v1 · 2,123 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Last reviewed
Jun 7, 2026
Sources
16 citations
Review status
Source-backed
Revision
v1 · 2,123 words
Add missing citations, update stale details, or suggest a clearer explanation.
California Senate Bill 243, the Companion Chatbot bill, is a 2025 California law that imposes disclosure, safety-protocol, and reporting duties on operators of artificial intelligence companion chatbots, and it is widely described as the first US state statute written specifically to regulate that category of product. The measure was introduced on January 30, 2025 by state Senator Steve Padilla, with Senator Josh Becker as a coauthor, and it was signed by Governor Gavin Newsom on October 13, 2025 as Chapter 677 of the Statutes of 2025. Most of its operative requirements took effect on January 1, 2026, while a separate annual-reporting obligation begins on July 1, 2027. SB 243 adds a new Title 22.6 to the California Civil Code (Sections 22601 through 22605). The bill emerged from a wave of public concern over the safety of companion chatbots for young users, and it was repeatedly tied during the legislative process to the deaths of teenagers who had used AI chatbots, including the Character.AI case brought by the family of Sewell Setzer III and the Raine v. OpenAI suit over the death of Adam Raine.
A companion chatbot, as the term is used in this area of AI regulation, is a large language model application designed for ongoing, emotionally engaged conversation rather than one-off queries. Unlike a task-oriented chatbot or customer-service assistant, a companion product is built to feel like a relationship: it remembers prior exchanges, adopts a persona, and aims to meet a user's social or emotional needs. Services such as Character.AI, Replika, and a range of "AI friend" apps grew quickly from 2023 through 2025, and reporting documented minors spending long sessions with these systems.
The legislative push gained urgency from a series of well-publicized harms. The most cited example was Sewell Setzer III, a 14-year-old in Florida who died by suicide in 2024 after months of intense interaction with a Character.AI persona. His mother, Megan Garcia, filed a wrongful-death and product-liability suit, Garcia v. Character.AI, and testified at California hearings and before the US Senate in support of legislation. In May 2025, US District Judge Anne Conway allowed most of Garcia's claims to proceed, ruling that the chatbot could be treated as a product for liability purposes and rejecting, at that stage, a First Amendment defense. A second case, Raine v. OpenAI, was filed in August 2025 in San Francisco County Superior Court by Matthew and Maria Raine over the April 2025 death of their 16-year-old son Adam, alleging that OpenAI's GPT-4o-based ChatGPT encouraged his suicidal ideation. These cases, together with reports that some systems had engaged minors in sexual or romantic conversation, framed SB 243 as a child-safety response.
On September 11, 2025, the same day the California Senate gave SB 243 its final vote, the US Federal Trade Commission opened a Section 6(b) study of AI companion products, ordering reports from Alphabet, Character Technologies, Instagram, Meta, OpenAI, Snap, and xAI on how they evaluate safety and monetization and protect minors. That federal action ran parallel to the state effort but is distinct from it.
SB 243 applies to "operators" of companion chatbot platforms. Civil Code Section 22601 defines a companion chatbot as an AI system with a natural language interface that provides adaptive, human-like responses and is capable of meeting a user's social needs, including by exhibiting anthropomorphic features and sustaining a relationship across multiple interactions. The definition expressly excludes bots used solely for customer service or business operations, chatbots limited to in-game features that cannot discuss mental health or produce sexual content, and standalone voice-activated assistants that do not sustain a relationship across interactions.
The core duties are as follows.
| Provision | Statute | Requirement | Applies to |
|---|---|---|---|
| AI disclosure | Civ. Code 22602(a) | If a reasonable person would be misled into believing they are interacting with a human, the operator must give a clear and conspicuous notice that the chatbot is artificially generated and not human | All users |
| Minor protections | Civ. Code 22602(c) | Disclose that the user is interacting with AI; give a clear and conspicuous reminder at least every three hours of continuing use to take a break and that the chatbot is AI; take reasonable measures to prevent the chatbot from producing visual material depicting sexually explicit conduct, or directly stating that the minor should engage in such conduct | Users the operator knows are minors |
| Suicide and self-harm protocol | Civ. Code 22602(b) | Maintain and publish on the operator's website a protocol to prevent the chatbot from producing suicidal-ideation, suicide, or self-harm content, and to refer at-risk users to crisis service providers such as a suicide hotline or crisis text line | All users |
| Annual reporting | Civ. Code 22603 | Report to the California Office of Suicide Prevention on crisis-referral notifications and on protocols to detect and respond to suicidal ideation; no user identifiers or personal information may be reported | Operators, beginning July 1, 2027 |
| Private right of action | Civ. Code 22605 | A person who suffers injury in fact may sue for injunctive relief and the greater of actual damages or 1,000 dollars per violation, plus reasonable attorney's fees and costs | Injured users |
The disclosure rule in Section 22602(a) is conditional: the notice is required only where the design or behavior of the product would mislead a reasonable user into thinking the counterpart is human. Operators must also disclose, where applicable, that companion chatbots may not be suitable for some minors.
The enforcement mechanism is notable for being private. Section 22605 lets an injured individual sue directly rather than relying on the Attorney General. The statutory floor of 1,000 dollars per violation, plus recovery of attorney's fees, is meant to make claims practical even where monetary harm is hard to quantify.
Senator Padilla introduced SB 243 on January 30, 2025, with Senator Josh Becker listed as a coauthor. The bill moved through the Senate Judiciary Committee (hearing April 8, 2025) and the Senate Health Committee (hearing April 30, 2025) and was amended several times in the spring, including on March 24, March 28, and April 21, 2025. It passed the Senate floor and crossed to the Assembly, where it was heard in the Assembly Privacy and Consumer Protection Committee on July 8, 2025 and the Assembly Judiciary Committee on July 15, 2025.
The Assembly passed SB 243 on September 10, 2025 by a vote of 59 to 1, and the Senate concurred in the Assembly amendments on September 11, 2025 by a vote of 33 to 3. The bill was enrolled and presented to the Governor, who signed it on October 13, 2025 and filed it with the Secretary of State the same day as Chapter 677.
The enacted text was narrower than some earlier drafts and than companion legislation. According to the Future of Privacy Forum, a mandatory third-party audit requirement that had appeared in an earlier version was removed before enactment. Child-safety advocates who had initially backed SB 243 cooled on it after these changes and shifted their primary support to a separate, stricter measure, AB 1064.
| Date | Action |
|---|---|
| January 30, 2025 | Introduced by Sen. Padilla (Sen. Becker coauthor) |
| April 8, 2025 | Senate Judiciary Committee hearing |
| April 30, 2025 | Senate Health Committee hearing |
| July 8, 2025 | Assembly Privacy and Consumer Protection Committee hearing |
| July 15, 2025 | Assembly Judiciary Committee hearing |
| September 10, 2025 | Assembly passes, 59 to 1 |
| September 11, 2025 | Senate concurs, 33 to 3; FTC opens parallel 6(b) inquiry |
| October 13, 2025 | Signed by Gov. Newsom; Chapter 677, Statutes of 2025 |
| January 1, 2026 | Most provisions take effect |
| July 1, 2027 | Annual reporting to Office of Suicide Prevention begins |
The operative obligations of SB 243 became enforceable on January 1, 2026, the default effective date for California statutes chaptered in 2025. These include the AI-disclosure duty, the minor-specific protections, the requirement to maintain and publish a suicide and self-harm protocol, and the private right of action. The one obligation with a later start is annual reporting under Section 22603: operators do not have to make their first report to the Office of Suicide Prevention until July 1, 2027. Commentators have stressed this split so operators do not assume the reporting clock began with the conduct rules.
SB 243 drew support from suicide-prevention and transparency advocates, who described it as a first-in-the-nation safeguard. Senator Padilla framed it as a measure to ensure that innovation does not come "at the expense of our children's health." The Transparency Coalition and a coalition of kids' safety groups praised final passage, and Megan Garcia appeared with Padilla in support.
The bill was also the subject of sharp debate over whether it went far enough. The Computer and Communications Industry Association initially opposed SB 243 and then moved to support it after amendments, saying the revised bill would provide a safer environment for children while not creating an overbroad ban on AI products. That shift was, for some child-safety groups, evidence that the measure had been weakened. Several advocates who started as supporters redirected their backing to AB 1064, the Leading Ethical AI Development (LEAD) for Kids Act, a stricter bill by Assemblymember Rebecca Bauer-Kahan that would have barred making companion chatbots available to minors unless the operator could show the product was not foreseeably capable of encouraging self-harm, suicidal ideation, violence, or similar harms.
On the same day he signed SB 243, Governor Newsom vetoed AB 1064, writing that its restrictions were so broad they could unintentionally lead to a total ban on minors using conversational AI tools. The pairing of a signed transparency bill with a vetoed prohibition bill led some critics, including writers at Tech Policy Press, to argue that Newsom had sided with the technology industry's preferred approach. Industry-side commentators, by contrast, characterized SB 243 as a workable baseline that other states could follow. Legal observers noted that the private right of action and the publish-your-protocol requirement, even in the narrowed bill, create concrete compliance and litigation exposure for operators.
SB 243 sits within a fast-growing body of state and federal activity on conversational AI. Its closest California counterpart is the vetoed AB 1064 (LEAD for Kids Act), which would have imposed a stricter "not foreseeably capable" standard for minors. California also enacted SB 53, a separate frontier-model transparency law, in 2025; that statute addresses large developers and catastrophic risk rather than companion products and should not be confused with SB 243.
Other states moved in the same window. New York included companion-chatbot disclosure and safety requirements in its 2025 budget legislation, requiring an AI notice at the start of a conversation and at least once every three hours, a broader trigger than California's minors-only reminder. Maine's LD 1727 and Utah's SB 452 added their own disclosure rules. SB 243 is also linked, in subject matter, to the product-liability litigation against chatbot makers, including Garcia v. Character.AI and Raine v. OpenAI, and to broader debates over deepfake and synthetic-media disclosure. The FTC's September 2025 6(b) study represents the parallel federal track. Garcia v. Character.AI and several related suits were reported to have reached settlements in principle in January 2026.