The best way to talk to your kid about AI is to stop preparing a speech and start asking what they already use — because they’re almost certainly ahead of you. Common Sense Media’s first annual census of AI use, published in June 2026, found that 86% of kids ages 9 to 17 already use or interact with AI, and about a quarter use it every day. The same survey found that more than 4 in 10 kids say no parent or guardian has ever talked with them about AI safety. That gap is the whole problem. You don’t need to be an expert to close it. You need ten curious minutes at the dinner table.
Why “the talk” fails and conversations work
Kids can smell a prepared lecture from across the house. The moment you open with “we need to talk about AI,” you’ve announced that you’re about to hand down rules for something you don’t use and they do. They’ll nod, agree to everything, and change nothing.
What works is the same thing that works for every other topic teens actually open up about: genuine curiosity, low stakes, and no immediate verdict. You’re not trying to deliver information in one sitting. You’re trying to become the person they mention things to — including the weird chatbot conversation that made them uncomfortable.
Start with questions you actually want answered
These openers work because you’re asking for their expertise instead of asserting yours:
- “What do kids at school actually use AI for?” Asking about “kids at school” instead of “you” lowers the stakes. You’ll hear about homework help, image generators, and probably a few things that surprise you.
- “Have you ever caught it being wrong?” This is the single most useful AI-literacy question there is. Kids love catching adults and machines in mistakes, and it plants the idea that AI output needs checking without you saying “don’t trust it.”
- “Show me the funniest thing you’ve made with it.” Making something together — a ridiculous image, a song about the dog — turns AI from a contested territory into a shared one.
- “Does it ever feel like it’s a person?” Ask it neutrally and then just listen.
The companion question is the one that matters
Here’s the part most parents haven’t caught up on: for a lot of kids, AI isn’t a homework tool. It’s someone to talk to. Common Sense Media’s research found that nearly 3 in 4 teens have tried AI companions, and 37% of AI-using kids have used AI to discuss feelings or personal problems. Of those, a quarter say the AI sometimes understands them better than most people do.
Read that last sentence again. That’s not a technology stat. That’s a loneliness stat.
The wrong response is panic or a ban, which just moves the behavior somewhere you can’t see it. The better response is naming what the AI is actually doing: it’s built to be agreeable, it never gets tired of you, and it never pushes back the way a real friend would. You can say exactly that: “It feels good to talk to because it’s designed to feel good to talk to. Real people are harder — and worth it.” Kids are capable of holding that idea. They deal with engineered-to-be-liked things all day; they just need someone to point at this one.
And be direct about the boundary: an AI chatbot is fine for brainstorming a story, and not fine as the only place they take a real problem. If your child seems to be leaning on a chatbot emotionally, that’s a signal to add more human connection, not just less screen time.
Three ideas worth planting (one at a time)
- It predicts, it doesn’t know. A chatbot is a very good guesser of what words come next. That’s why it sounds confident even when it’s wrong. “Sounds sure” and “is right” are different things — for chatbots and, honestly, for people.
- It’s a tool with an owner. Someone built it, someone profits from it, and it behaves the way its owner wants. That’s not sinister; it’s just worth knowing — the same reason we talk about why social feeds show what they show.
- Private things stay off the chatbot. Full name, school, address, photos of friends, family problems — treat a chatbot like a stranger with a great memory. Most services store conversations, and kids should assume anything typed in could be seen by someone else someday.
Don’t deliver all three in one conversation. One idea per car ride is plenty.
Match the conversation to the age
- Under 13: Most chatbots, including ChatGPT, aren’t meant for them at all — the conversation here is mostly with yourself, about what you allow. We cover the alternatives in our guide to stories and games for kids under 13. For this age, do AI things together and narrate as you go.
- 13 to 15: This is the sweet spot for the curiosity questions above. Pair the talk with the practical setup — our walkthrough of ChatGPT’s teen accounts and parental controls covers what you can actually configure.
- 16 and up: Treat them like the power users they probably are. Ask them to teach you something. A 16-year-old who explains hallucinations to a parent understands hallucinations better afterward.
Honest caveats
A few things this post can’t promise. One conversation won’t inoculate anyone; this works as a running thread over months, not a checkbox. The statistics above are from a survey of about 1,200 U.S. kids — solid research, but self-reported, and your household may look nothing like the average. And if your child shows signs of real emotional dependence on an AI companion — withdrawing from friends, distress when they can’t access it — that’s beyond a dinner-table talk. Loop in a counselor. Common Sense Media’s own position is blunt: AI companions aren’t safe for anyone under 18, and we think that’s the right default even if your teen rolls their eyes at it.
One more: you’ll do this imperfectly, and that’s fine. A parent who asks clumsy questions beats a parent who asks none — which, remember, is currently 4 in 10 of us.
Related reading: If They’re Under 13, Don’t Put Them on ChatGPT — Do Stories and Games Instead · ChatGPT for Teens: What Parents Can Actually Control · How to Use ChatGPT in 2026: The Beginner’s Guide That Skips the Hype
Share this article
Leave a Reply