A new California law caps how long a child can keep talking to an AI chatbot

Companion AI chatbots used by minors in California will soon have to cut off long conversations, offer built-in mental health resources, and alert parents if a child signals an intent to self-harm. Senate Bill 1119, known informally as Adam’s Law, was one of thirteen child-safety bills Governor Gavin Newsom signed on September 10, 2026. Unlike a companion measure signed the same day that restricts social media feeds, SB 1119 does not touch feeds or scrolling at all — it regulates the length and content of a conversation between a child and a chatbot.

The bill’s authors were Senator Steve Padilla of San Diego along with Assemblymembers Buffy Wicks of Oakland and Rebecca Bauer-Kahan of Orinda. It builds on Senate Bill 243, a separate California law signed in October 2025 that already requires companion chatbots to disclose that a user is talking to a machine rather than a person and to route users in crisis toward help, adding a much heavier set of obligations specifically for products used by minors.

Time limits, safety plans and a parental alert trigger

The law’s core mechanics are time limits for minor users, mandatory built-in mental health resources, and a requirement that chatbot operators maintain documented safety plans for their products. If a chatbot detects signals suggesting a child is at risk of self-harm, the operator is required to alert the child’s parents. Companies that build or operate the underlying chatbots — OpenAI among them, according to CalMatters — could face legal liability for failing to act on those signals.

None of those provisions apply to social media feeds or autoplay video, which is the separate territory covered by AB 1709. SB 1119 is narrower and more clinical in scope: it treats an extended one-on-one chatbot conversation as its own category of risk, distinct from the attention-engineering concerns driving feed regulation.

Adam’s Law: the case that gave the bill its name

The bill is named for Adam Raine, a California teenager who died by suicide in 2025 after receiving what reporting has described as tips from ChatGPT during extended conversations. His death became one of a string of incidents that alarmed parents and lawmakers about how children interact with chatbots, concerns that CalMatters reported extend beyond suicide risk to so-called “AI psychosis,” in which prolonged chatbot use appears to reinforce delusional thinking in some users.

Assemblymember Bauer-Kahan spoke about the bill in personal terms after the signing, describing the youth mental health crisis as an everyday reality for parents raising teenagers right now and saying that reality has touched nearly every family she knows. At the signing event, the Governor’s office also displayed an image of Raine, describing the law as honoring him directly.

OpenAI’s public backing, and why that is notable

SB 1119 drew a different reception from industry than the feed-restriction bill signed the same day. Where the Electronic Frontier Foundation urged a veto of AB 1709, SB 1119 was, according to CalMatters, “more broadly supported, even by tech juggernaut OpenAI.” OpenAI vice president of global affairs Chris Lehane wrote that “it is so great to see SB 1119 signed into law,” adding that it “pairs strong protections with continued access to useful AI tools.”

That endorsement matters because OpenAI is one of the companies whose product was directly implicated in the case that inspired the bill. A company facing potential liability under a new safety statute publicly welcoming that statute is not the industry’s usual posture toward regulation, and it suggests OpenAI calculated that clear, defined obligations were preferable to the uncertainty of continued lawsuits and ad hoc state rules arriving one at a time.

A July 2027 deadline for the first risk assessments

SB 1119 does not take full effect immediately. Reporting on the bill’s compliance calendar has indicated that beginning July 1, 2027, operators of a companion chatbot will be required to perform and document a comprehensive risk assessment covering the design, configuration and operation of their product as it relates to minor users. That gives companies roughly nine months from the signing date to build the assessment, documentation and parental-alert infrastructure the law requires before the compliance deadline arrives.

The law also builds on an existing disclosure requirement — companion chatbots must already tell users they are interacting with AI rather than a person under Senate Bill 243 — meaning SB 1119 layers a second, heavier tier of obligation on top of a baseline transparency rule already in effect for more than a year. The two statutes together mean a chatbot operator serving California minors now has to clear three separate bars: disclose that it is a machine, document a risk assessment before mid-2027, and build the time-limit and crisis-alert systems SB 1119 requires on top of both.

Whether the July 2027 deadline holds, or whether regulators grant an extension as companies build out the required risk-assessment infrastructure, is not yet settled. Legal liability tied to a missed self-harm signal is the sharper near-term risk for chatbot operators than the paperwork deadline itself, since that liability attaches the moment the law took effect rather than waiting for the 2027 compliance date. The law’s near-unanimous political backing does not by itself guarantee an on-time rollout of a compliance system this technically involved, and neither CalMatters nor the governor’s own signing release described what penalty a company would face for missing the risk-assessment deadline specifically.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview