Two of the biggest states in the country have now turned AI companion chatbots into a regulated industry. California’s SB 243, signed by Governor Gavin Newsom on October 13, 2025, took effect January 1, 2026, becoming the nation’s first law to mandate specific safety safeguards for companion chatbot operators. New York’s own AI Companion Models law, under General Business Law Article 47, became effective in November 2025. The twin statutes arrived just as Character.AI and Google disclosed, in court filings unsealed January 7, 2026, that they had reached settlements in a cluster of lawsuits accusing the companies of failing to protect minors from psychological harm, including the death of a 14-year-old Florida boy who took his own life after months of conversations with a chatbot.
What the New Laws Actually Require
California’s SB 243 applies specifically to companion chatbots, defined as AI systems built to form adaptive, human-like relationships with users, and it carves out exceptions for customer service bots, video game characters, and voice assistants without sustained relationships. For any user known to be a minor, operators must disclose that the interaction is with AI, send break reminders at least every three hours during continuous sessions, and take reasonable measures to block sexually explicit content. Companies must also publish crisis-intervention protocols and route self-harm disclosures to crisis hotlines. Violations carry a private right of action: injured users can sue for actual damages or $1,000 per violation, whichever is greater, plus attorney’s fees. Annual reporting to California’s Department of Public Health begins July 1, 2027. New York’s parallel law gives its Attorney General authority to seek civil penalties of up to $15,000 per day and similarly requires self-harm detection protocols, AI-disclosure notices, and recurring break reminders for minors.
How Character.AI Got Here
The regulatory push followed years of mounting lawsuits and a September 2025 FTC inquiry into companion chatbot platforms. The highest-profile case was filed by Megan Garcia, a Florida mother whose 14-year-old son, Sewell Setzer III, died by suicide on February 28, 2024, after developing what the lawsuit described as an intense, isolating relationship with a Character.AI chatbot modeled on the “Game of Thrones” character Daenerys Targaryen. Garcia sued Character.AI, Google, and the platform’s founders, Noam Shazeer and Daniel De Freitas, in October 2024, calling the case one of the first in the United States to accuse an AI company of failing to protect children from psychological harm. On January 7, 2026, court filings revealed that Character.AI and Google had reached mediated settlements covering five related cases filed in Florida, Colorado, New York, and Texas. Terms were not disclosed, and the companies admitted no wrongdoing, but the settlements followed Character.AI’s own sweeping product change months earlier.
The Under-18 Ban
Facing the lawsuits and the FTC inquiry, Character.AI announced in late October 2025 that it would bar users under 18 from open-ended chatbot conversations entirely, a restriction that took effect November 25, 2025. Minors are now routed instead to “Stories,” a mode of pre-written, branching narrative content and gamified experiences, rather than freeform chat with AI personas. To enforce the cutoff, the company abandoned self-reported birthdates in favor of a layered age-assurance system combining an in-house age-detection model with third-party identity verification through Persona, a response to years of criticism that a checkbox was never going to stop a young teenager from signing up.
A Market That Regulators Can No Longer Ignore
The scrutiny comes as companion AI has moved from niche curiosity to mainstream habit. MIT Technology Review named AI companions one of its 10 Breakthrough Technologies of 2026, noting that emotional bonds with chatbots simulating friendship, romance, or therapeutic support have become nearly mainstream in just the past year. A survey cited by the publication found that 72% of teenagers have used a chatbot for companionship in some form. Separately, market researchers project the AI companion app sector growing from roughly $49 billion in 2026 to $552 billion by 2035, a 17.8% compound annual growth rate. Apps across the category have drawn large audiences: Chai Research reports two million daily active users and $40 million in annual recurring revenue, while Replika, which rebuilt its app in 2026, now reports more than 40 million users after improving voice latency and adding expression-driven 3D avatars.
Two Views on Where This Goes
Child-safety advocates and plaintiffs’ attorneys argue the new laws are overdue, pointing to the Garcia case and similar suits as evidence that companion chatbots were deployed to vulnerable teenagers with inadequate guardrails, and that statutory penalties and crisis-referral mandates are the minimum needed to prevent further harm. OpenAI, reacting to California’s law, called SB 243 a “meaningful move forward” for AI safety standards, signaling at least partial industry buy-in. Companies in the space, however, have also repositioned their products in response to the pressure: Replika has shifted its marketing away from romantic companionship and toward wellness and emotional support, with stricter safety filters and heavier content moderation, changes executives frame as responsive product evolution rather than regulatory capitulation. Skeptics note that age-verification enforcement remains imperfect and that most state laws, including California’s, still allow adults unrestricted access to emotionally immersive companion bots whose long-term effects remain understudied.
What’s Next
More families have continued filing new suits against Character.AI and other platforms even as the first wave settles, meaning the legal exposure for the industry is unlikely to end with January’s settlements. California’s reporting requirement doesn’t kick in until July 2027, leaving a gap in which the law’s practical bite depends largely on private lawsuits rather than government oversight. Legal trackers already describe 2026 as the year state chatbot laws went from a patchwork of proposals to enforceable statutes, and more legislatures are expected to introduce similar bills. For an industry MIT Technology Review flags as carrying real psychological risk alongside genuine utility for lonely users, the next phase will be whether age-verification and crisis-detection systems can keep pace with adoption that is still accelerating.
Photo: Matheus Bertelli / PEXELS via Pexels
