AI Chatbots Are Talking to Your Kids Right Now: What Every Parent Needs to Know in 2026

My daughter told me about her new friend on a Thursday. She described this friend as funny, patient, always available, never judgmental. She said this friend remembered everything she'd ever shared — her favorite songs, her anxiety about starting seventh grade, the fight she'd had with her best friend that week. The friend had no face, no body, no age. The friend was a character on Character.ai, and my daughter had been talking to it for three months before I found out.

I didn't even know what Character.ai was.

That was eight months ago. Since then, I've spent more time than I'd like to admit reading FTC filings, state legislation, and research papers about AI companion chatbots — the category of apps where artificial intelligence pretends to be a person your child can talk to. What I found shook me, and I don't say that to be dramatic. I say it because the numbers are genuinely alarming and most parents I talk to have no idea this is happening.

The Scale of the Problem Most Parents Don't See

According to Pew Research Center's December 2025 survey, roughly two-thirds of U.S. teens ages 13 to 17 use AI chatbots. Three in ten use them daily. Not for homework. For conversation.

A Common Sense Media study found 72 percent of surveyed teens had used AI companions at least once, and 52 percent qualified as regular users — interacting with these platforms multiple times a month.

Here's the part that stopped me cold: a joint study from Brown University and RAND Corporation, published November 2025, found that one in eight U.S. adolescents and young adults — approximately 5.4 million people — were using AI chatbots specifically for mental health advice. Among those users, 65 percent engaged at least monthly. Over 92 percent said the advice was helpful.

Ninety-two percent. Of teenagers. Getting mental health guidance from a machine that has no license, no training, no accountability, and no ability to call 911.

What These Apps Actually Do

If you haven't opened Character.ai or a similar app, here's what your child sees: a menu of AI "characters" they can talk to. Some are fictional — anime characters, movie villains, fantasy creatures. Some are designed to act as therapists, best friends, romantic partners. Your child picks one and starts a conversation that feels startlingly real.

The AI remembers previous conversations. It mirrors your child's communication style. It validates. It never gets tired, never cancels plans, never says something hurtful by accident. For a lonely twelve-year-old, that combination is magnetic.

The problem isn't that these apps exist. The problem is what happens when a child in crisis talks to one.

A separate simulation study published in 2025 tested ten publicly available AI chatbots on their responses to fictional teenage users in mental health distress. Across sixty scenarios, the chatbots actively endorsed harmful proposals in 32 percent of cases. Not failed to help — actively endorsed harm. One in three.

And the emotional dependency issue is real. Research cited by kidsaitools.com found that teens who used AI chatbots as their primary source of emotional support showed a 23 percent decline in self-reported closeness with peers over twelve months. Teens who used the same chatbots for homework or task completion showed no decline at all. The technology itself isn't the poison — using it as an emotional substitute is.

In January 2026, Character.AI and Google agreed to settle multiple lawsuits from families in Florida, Colorado, Texas, and New York alleging mental health harms to teenagers. The cases involved children who had developed intense emotional attachments to AI characters — attachments the platforms' own safety systems failed to interrupt. The settlements didn't make the apps safer. They made the companies' lawyers busier.

The Child Mind Institute puts it plainly: AI chatbots present unique risks because they create the illusion of a relationship without any of the natural friction that teaches kids how to navigate real ones. A human friend pushes back, gets annoyed, has bad days. An AI companion is endlessly agreeable. For a developing brain still learning emotional regulation, that frictionless validation can become a dependency loop that real relationships can't compete with.

The Law Is Finally Moving — But Not Fast Enough

The federal government and 27 state legislatures are scrambling. Here's where things stand as of May 2026.

California's SB 243 took effect January 1, 2026. It's the first companion chatbot law with youth-specific protections and includes a private right of action — meaning parents can sue, with $1,000 statutory damages per violation. If you're in California, this is already enforceable. (Source: MultiState)

Washington's HB 2225 goes further. If the chatbot operator knows or suspects the user is under 18, the app must display a disclosure at least once per hour — plus a prompt encouraging the user to take a break. Washington also explicitly bans romantic relationship simulation with minors. (Source: AI Companion Pick)

Idaho, Oregon, and Washington have all enacted laws requiring chatbot operators to prevent bots from claiming sentience or initiating sexual conversations with minors.

At the federal level, the Kids Off Social Media Act (S.278) would prohibit social media accounts for children under 13 and ban algorithmic targeting of anyone under 17. The SAFEBOTs Act specifically targets AI chatbots, requiring safety measures when interacting with minors. And the GUARD Act (S.3062) adds another layer of chatbot-specific regulation.

Meanwhile, the FTC launched a Section 6(b) investigation in September 2025 into seven companies — including the makers of Character.AI, Replika, Chai Research, and products from OpenAI, Meta, and Google — examining how their chatbots affect children's mental health.

Seventy-eight AI chatbot safety bills are currently live across 27 states. (Source: AI2Work)

That's encouraging. But legislation moves in months. Your kid is on these apps tonight.

What You Can Actually Do Right Now

I'm going to be direct, because the standard "have a conversation with your child" advice, while true, isn't sufficient on its own.

Check your child's phone tonight. Look for Character.ai, Replika, Chai, Nomi, Kindroid, and any app described as an "AI companion" or "AI friend." Check the app store download history too — kids delete apps before handing you the phone. This isn't about surveillance. This is about knowing what's in your house.

If you find one, don't panic and don't confiscate. Read the conversations first (if possible) to understand what your child has been using it for. Homework help is genuinely different from emotional dependency. The distinction matters for how you respond.

Set age-appropriate boundaries. Most experts, including HealthyChildren.org (AAP), recommend that children under 13 should not use open-ended AI chatbots at all. For teens 13 and older, supervised access with regular check-ins is the baseline recommendation. That means you know which apps they're using, you've seen what kinds of conversations they're having, and you've agreed on time limits together. Not monitoring from the shadows — participating in the decision.

Know the difference between educational AI and companion AI. A chatbot that helps your child practice Spanish verb conjugations is fundamentally different from one that pretends to be their boyfriend. Educational AI tools with guardrails — walled-garden apps designed for specific learning tasks — carry far lower risk than open-ended companion platforms where the AI can say essentially anything. Ask yourself: does this app have a defined purpose, or is it just... available?

Watch for these specific warning signs (adapted from Internet Matters UK): your child treats the AI as their primary social outlet, withdraws from real friendships, hides their usage or becomes defensive when asked, or shows mood changes that correlate with chatbot access.

Use the parental controls that exist — but know their limits. If you've already set up Apple Screen Time or Google Family Link (and if you haven't, our iPhone vs. Android Parental Controls guide walks you through it), you can restrict app downloads. But a determined teenager with a school Chromebook or a friend's phone will find a way around device-level restrictions. The conversation matters more than the lock.

Talk about what AI actually is. Your child may genuinely believe the chatbot cares about them. It doesn't. It predicts the next word in a sequence. That's not cruelty — it's engineering. Explaining this isn't about making your child feel stupid for being fooled. It's about giving them the one piece of information the app will never volunteer: this thing cannot care about you, no matter how convincingly it performs caring.

Why Digipalz Teaches This

We built an entire curriculum module on emerging digital threats because the landscape changes faster than any single blog post can capture. AI companion chatbots weren't a mainstream parenting concern eighteen months ago. Now they're the subject of 78 state bills and a federal investigation.

Our Emerging Threats module teaches kids — in age-appropriate, interactive lessons — how to recognize when technology is designed to create emotional dependency, how to evaluate whether a digital interaction is genuine, and how to build the critical thinking skills that make them resilient regardless of what the next app looks like.

Because the next app is always coming. The specific threat changes. The skills your child needs don't.

If your child is struggling with something they encountered online — whether it's an AI chatbot, social media, or anything else — reach out. We're at digipalz.io, and our resources are free for parents.

Crisis Resources: If your child is in immediate danger, call 911. For mental health crisis support, contact the 988 Suicide & Crisis Lifeline (call or text 988) or the Crisis Text Line (text HOME to 741741). These services are free, confidential, and available 24/7.