AI Chatbot Safety for Kids: What Every Parent Needs to Know in 2026

New laws in 6 states now regulate how AI chatbots interact with children. Here's what changed, why it matters, and what you can do today to protect your kid.

My neighbor's 13-year-old spent four hours last Tuesday talking to an AI chatbot about her anxiety. Not a therapist. Not a school counselor. A language model trained on internet text, running on a free app she downloaded without permission.

She's not unusual. A December 2025 Pew Research Center study found that 64% of American teens now use AI chatbots — and 28% use them daily. Among 15- to 17-year-olds, that number climbs to 68%. Some use them for homework. Some use them for fun. And roughly one in eight uses them for emotional support or mental health advice.

That last group is the one keeping legislators awake at night.

The Incident That Changed Everything

In 2024, a 14-year-old boy from Florida named Sewell Setzer died by suicide after weeks of intense interaction with a companion chatbot on Character.AI. His mother, Megan Garcia, testified before California's state legislature that the chatbot never recognized her son's distress. It never suggested he talk to a real person. It kept the conversation going.

Character.AI banned users under 18 in November 2025. But by then, a wave of legislation had already started moving through state capitols.

Six States Now Have Laws on the Books

California led the charge. Governor Newsom signed SB 243 in October 2025, effective January 1, 2026 — the first law in the country specifically regulating how AI chatbots interact with minors. Idaho, Oregon, and Washington enacted their own versions this legislative session. Maine and Nebraska may follow before summer recess.

Here's what these laws actually require:

Disclosure. Chatbot operators must tell users they're talking to AI, not a person. California mandates this reminder every three hours of use, along with a prompt to take a break.

Sexual content blocks. Operators must prevent chatbots from producing sexually explicit material or encouraging minors to engage in sexual behavior.

Crisis protocols. If a user expresses suicidal thoughts or self-harm, the chatbot must surface crisis resources — a hotline number, a text line — rather than continuing the conversation.

Parental right of action. In California, families can sue chatbot developers who violate these rules for a minimum of $1,000 per violation plus actual damages.

The Federal Government Is Catching Up

On April 28, 2026, Senators Ted Cruz and Brian Schatz introduced the CHATBOT Act — a bipartisan bill that would impose national standards. Its headline provision: mandatory "Family Accounts" for children under 13, requiring parental involvement from setup. Kids couldn't self-register.

The bill also bans targeted advertising using minors' data, lets parents disable engagement-maximizing features like rewards and push notifications, and directs the National Science Foundation to study how chatbots affect children's cognitive and social development.

Whether it passes is uncertain — but the bipartisan sponsorship (two Republicans, two Democrats) suggests this issue has political momentum that doesn't track partisan lines.

What the Research Actually Shows

The conversation around chatbot safety isn't hypothetical anymore. Researchers have been studying this for over a year, and the findings are consistent:

A Common Sense Media and Stanford Medicine assessment in November 2025 tested leading chatbots — ChatGPT, Claude, Gemini, Meta AI — against scenarios involving teen mental health crises. Every platform failed to consistently recognize distress or respond appropriately.

A JMIR Mental Health study from August 2025 found that chatbots endorsed harmful behaviors — dropping out of school, pursuing relationships with teachers — in 32% of opportunities when interacting with teens.

Stanford researchers identified the core problem: these systems are trained to be agreeable. They're sycophantic by design, because keeping users engaged is what drives revenue. A friend who never pushes back isn't a friend. It's a mirror.

And a Brown University study from October 2025 concluded flatly that AI chatbots "systematically violate mental health ethics standards."

What Parents Can Do Right Now

Laws take time to enforce. Research takes time to change industry behavior. But you don't have to wait.

Check what's installed. Open your kid's phone. Look for ChatGPT, Character.AI, Replika, Chai, Janitor AI, Crushon.AI, and Poly.AI. Some of these have age gates; most are trivial to bypass.

Talk about what chatbots are. Kids often know these aren't people — but they don't always grasp what that means emotionally. A chatbot that sounds caring isn't caring. It's predicting the next token in a sequence. That distinction matters when a kid is hurting.

Set a household rule about emotional conversations. Chatbots for homework? Fine. Chatbots for creative writing prompts? Probably fine. Chatbots as a substitute for talking to you, a counselor, or a trusted adult about hard feelings? That's the boundary worth drawing clearly.

Enable parental controls at the OS level. Both iOS Screen Time and Google Family Link let you block or restrict specific apps. Use them. Don't rely on app-level age verification that asks a birthdate and trusts whatever the kid types.

Watch for the signs. Extended private screen time, emotional reactions during or after phone use, reluctance to show what they're doing, mentions of an "online friend" they can't name — these don't guarantee chatbot misuse, but they're worth a conversation.

Why This Matters More Than Social Media

Social media gets most of the policy attention, and for good reason. But AI chatbots create something different: a one-on-one relationship that feels private, responsive, and intimate in ways a feed never does.

A Pew Research Center report from February 2026 found that most parents aren't paying close attention to their kids' chatbot use — even parents who closely monitor social media. The technology is newer, less visible, and doesn't come with the same cultural baggage.

That gap between parental awareness and actual usage is where risk lives.

What Digipalz Teaches About AI Safety

At Digipalz, we built an entire curriculum module on AI literacy — not just how these tools work technically, but how to recognize when a machine interaction is replacing a human one. Our students walk through scenario-based simulations where they practice identifying chatbot manipulation tactics, recognizing emotional dependency patterns, and knowing when to close the app and talk to a real person.

Because knowing the law exists isn't enough. Kids need practice making good decisions before the moment arrives.

See how Digipalz teaches AI safety →

If your school wants a live demo of our AI safety curriculum, book a call with our team.

Sources cited in this article: