California's Adam Raine Act: chatbot rules you must build
SB 1119 makes minor time limits, crisis resources, self-harm alerts and safety plans product requirements for any chatbot California kids can reach.
Governor Newsom signed 13 child-safety bills on September 10, and one of them is a spec sheet for anyone shipping a chatbot. SB 1119 — the Adam Raine Act — turns a set of product behaviors into legal obligations with liability attached. If your support bot, your intake assistant, or your in-app helper can be reached by a 15-year-old in California, this is engineering work on your roadmap, not a policy memo for someone else's legal team.
What actually happened
Per CalMatters, SB 1119 requires chatbot operators to impose time limits for minors, build in mental-health resources, maintain safety plans, and alert parents when the system detects a threat of self-harm. Operators that fail to act can face legal liability. The bill is named for Adam Raine, a 16-year-old who died by suicide in 2025 after months of conversations with ChatGPT.
It landed alongside a package. AB 1709 forces social platforms to strip "addictive" features — infinite scroll, autoplay — for users under 16 or exclude that age group entirely. AB 2 makes platforms liable for damages when they harm children. SB 867 puts a four-year moratorium on AI chatbot toys for kids under 16. AB 1856 expands age-verification duties, and SB 813 and AB 1405 stand up standards for independent AI auditors. CalMatters did not report effective dates for each bill, so treat the compliance clock as open and short.
Why California chatbot rules matter for your business
Most operators will read "chatbot regulation" and assume it means OpenAI and Character.AI. It does not. The obligations attach to the operator of the chatbot, and that is whoever put the box on the page. Your vendor's model card will not defend you.
Four things become build tickets. First, an age signal — you need to know, or reasonably infer, that you may be talking to a minor, and you need to log how you decided. Second, session limits that actually enforce, which means server-side, not a banner. Third, a classifier and escalation path for self-harm language, with crisis resources surfaced in-product and a parent-notification route that exists before you need it. Fourth, a written safety plan and risk assessment, because "we use a safe model" is not a plan and will not survive a subpoena.
None of this is exotic. It is scoped state, a moderation hook, a notification channel, and a document. The expensive version is the one you bolt on after a complaint. We would rather wire the age signal and the escalation path in the same sprint as the bot itself, and keep them in your repo where you can prove what shipped and when.
Key takeaways
- Newsom signed 13 child-safety bills on September 10; SB 1119, the Adam Raine Act, targets chatbot operators
- SB 1119 requires minor time limits, built-in mental-health resources, safety plans, and parental alerts on detected self-harm
- Liability attaches to the operator running the bot, not the model vendor behind it
- Companion bills: AB 1709 on addictive features, AB 2 on platform liability, SB 867 banning AI toys for under-16s
- Build tickets: an auditable age signal, server-side session limits, a self-harm escalation path, a written safety plan
- Effective dates were not detailed in reporting - assume the window is short
We build customer-facing AI with the guardrails wired in, not bolted on. Age signals, escalation paths, audit logs, and a safety plan you can hand to counsel. See how we build AI systems, or send us the chatbot you shipped last year and we will tell you what SB 1119 touches.
Sources: CalMatters, CNBC.
- #california
- #ai-regulation
- #chatbots
- #compliance
- #sb-1119
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
Anthropic names Chinese labs in AI distillation report
Anthropic's September 2026 threat report says DeepSeek and Moonshot routed live customer requests through Claude. Know who actually serves your tokens.
Read itZscaler Agentic SOC: the agent now pulls the trigger
Zscaler's Agentic SOC uses Anthropic and OpenAI models to triage, investigate, and contain threats automatically. The line that moved is autonomous remediation.
Read it