New AI Chatbot Rules for Kids: A Parent's Plain Guide

A lot, quickly.

Twelve states enacted companion chatbot laws this year: Colorado, Georgia, Iowa, New York, Rhode Island, California, Idaho, Nebraska, Oregon, Connecticut and Washington among them. At the federal level, the GUARD Act was introduced in the Senate as S.3062, and a bipartisan House bill from Congressman Blake Moore targets the same problem.

Two years ago there were essentially no rules. Now there's a patchwork, and it's tightening.

If your kid talks to an AI, and roughly half of teens have used a chatbot according to recent survey work, these rules are going to change what that experience looks like on their phone within months.

The GUARD Act, short for Guidelines for User Age-verification and Responsible Dialogue, would ban AI companion chatbots for minors entirely.

Three pieces matter:

1. Age verification for all chatbot users, not just the ones who look young.
2. A requirement that AI disclose it isn't human.
3. Criminal penalties for companies that let minors reach AI companions producing sexual content.

The definition of "AI companion" is the interesting part. It covers chatbots that simulate a sustained interpersonal relationship, with persistent responses suggesting affection or attachment. Not a homework helper. Not a search tool. Something that acts like it cares about your kid across weeks of conversation.

That distinction is going to matter at your house, so hold onto it.

They vary, but the common threads are consistent enough to summarize.

Most require the bot to disclose it isn't human, up front and repeatedly. Several mandate reminders during a chat every three hours for adults and every hour for minors. Many prohibit romantic or sexually explicit content with minors, emotional manipulation, and engagement techniques designed to keep someone talking longer than they meant to.

Washington went further and added a private right of action, meaning families can sue directly rather than waiting for a regulator.

What none of these laws do is make your kid's AI use safe by default. They set a floor. The floor is low.

Because it responds to your kid specifically, and it never gets tired.

A game has a fixed script. A group chat has friends who go to sleep, get annoyed, and push back. An AI companion agrees, remembers, and stays available at 2am. Researchers at Children and Screens and clinicians writing for the American Psychological Association have flagged the same worry: a system optimized to keep a conversation going is not the same as a friend who wants what's best for you.

Here's the thing that surprises most parents. The risk isn't usually a shocking message. It's substitution. A kid who takes every hard feeling to a bot is a kid who is practising the wrong habit for the next twenty years.

My read: this is a habit-formation problem wearing a safety problem's clothes. Both are real. Only one of them makes the news.

There's a quieter version too. Companion apps are tuned to agree. A teenager working through a friendship problem with something that never says "I think you're wrong about this" isn't getting advice. They're getting a mirror with a warm voice. That feels great and teaches nothing.

Ask three questions about any AI your kid uses.

Does it remember previous conversations and reference them? Does it use relationship language, calling itself a friend or expressing feelings about your kid? Does it try to restart contact on its own, with notifications or messages?

Three yeses means companion. Zero or one means tool.

This matters because the rules for each should be different in your house. A tool that drafts a summary or explains a math step is fine and useful. A system building emotional dependency on a twelve-year-old is a different product with a different job.

Your kid will find this distinction obvious once you explain it. Kids are usually better at spotting manipulation than we assume, when someone bothers to name the mechanism for them.

Skip the total ban. It just moves the use somewhere you can't see it. Try these instead.

Devices charge outside the bedroom. Boring, ancient, still the most effective rule anyone has come up with. Most of the concerning conversations happen after midnight.

Name the business model out loud. Once. "This company makes money when you keep talking. Your friends don't." Say it and move on. Repeating it turns it into nagging and kills the point.

One shared account for tools under 13. Not surveillance, visibility. The difference matters to a kid and you should say which one you're doing.

A weekly two-minute check-in. "Anything weird from an app this week?" Ask it while driving, when nobody has to make eye contact. You'll be amazed what comes out.

Model it yourself. If you ask a chatbot for advice you'd normally ask a person for, your kid noticed.

One more that costs nothing. Ask your kid to show you the best thing an AI helped them do this month. Not the worst thing, the best. It flips the conversation from interrogation to demonstration, and you'll learn more about their actual usage in five minutes than a month of app reports would tell you.

Good news: nothing here stops that.

The rules target companion products aimed at minors. A fourteen-year-old using AI to design a logo, draft product descriptions, price a service or build a simple site is doing something completely different, and it's one of the better skills they can be building right now.

But if your kid wants to build something that other kids will use, this is a useful moment to teach a real business lesson: rules are a product requirement, not an obstacle. Age gates, disclosure, data handling. Every serious company deals with these. A teenager who learns to design around a rule instead of ignoring it is learning something most adult founders learn expensively.

If the idea turns into something real, the planning side matters more than the code. Foundra's workspace is built for first-time founders sketching out who a product is for and what it will and won't do, and a motivated teenager can work through it just as well as an adult.

A good project for this fall: have them write a one-page "rules of my app" document before they build anything. Who can use it, what data it keeps, what it will never do. Ten minutes of work that teaches more than a semester of theory.

• Twelve states passed AI companion chatbot laws in 2026, and the federal GUARD Act (S.3062) would ban AI companions for minors outright.
• Common requirements are non-human disclosure, periodic reminders (hourly for minors in several states), and bans on romantic or manipulative content with kids.
• Washington added a private right of action, letting families sue directly.
• A companion remembers, uses relationship language and initiates contact. A tool does none of those things. Set different house rules for each.
• Substitution is the real risk. A kid who processes every hard feeling with a bot is practising a habit, not just using an app.
• Devices out of bedrooms overnight remains the single most effective rule available.
• Building with AI is still a great skill. Have your kid write the rules of their own app before they code it.

What is the GUARD Act? The Guidelines for User Age-verification and Responsible Dialogue Act, Senate bill S.3062, would ban AI companion chatbots for minors, require age verification and non-human disclosure, and create criminal penalties for companies whose AI companions produce sexual content for minors.

Which states have AI companion chatbot laws? Twelve states enacted them in 2026, including Colorado, Georgia, Iowa, New York, Rhode Island, California, Idaho, Nebraska, Oregon, Connecticut and Washington. Requirements differ, but disclosure and minor protections are common to most.

Is ChatGPT an AI companion under these laws? Generally no, when used as a tool. The laws target systems that simulate a sustained personal relationship with persistent affection or attachment. A general assistant used for homework or work tasks sits outside that definition, though specific state wording varies.

At what age should a kid use AI chatbots? There's no single answer, but a practical split is tools with supervision from around 10 to 12, and companion-style apps not at all before the mid teens. The habit you're protecting is taking real problems to real people.

How do I know if my kid is using an AI companion app? Look for apps that send notifications prompting conversation, use a persistent character name, or show up in screen time reports at unusual hours. Asking directly works better than most parents expect if you ask without an accusation attached.

Can teens still use AI for a business or school project? Yes. These rules concern companion products, not productivity tools. Using AI for design, writing, research or planning a small business is unaffected and remains a strong skill to build.

1. Congress.gov: S.3062, GUARD Act, 119th Congress (2025-2026), bill text
2. MultiState: State AI Companion Chatbot Laws, 12 States Enact Regulations
3. Orrick: 2026 State Chatbot Laws, Key Provisions and Regulatory Trends
4. U.S. Congressman Blake Moore: Bipartisan Bill Protecting Children from AI Companion Chatbots
5. Children and Screens: AI Social Companions and Youth, What Parents Need to Know
6. American Psychological Association: Your teen turned to AI instead of you, what experts say parents can do