BloomPath exists because parenting keeps handing us problems the last generation never had to Google. Last month, at pickup at Luna’s Montessori school, another dad — I’ll call him Mike, because that’s actually his name and he said it was fine — mentioned that his son had been talking to an AI character every night before bed. Not gaming with it. Talking to it. About his day, his worries, a fight with a friend.

Mike wasn’t panicking. He said it almost like a funny aside: “At least he’s not asking me for once.” I laughed too, for about four seconds. Then I went home and looked up what his son was actually using, and I stopped laughing.

TL;DR: AI companion chatbots — apps designed to feel like a friend, not a search engine — are now used by a large share of kids and teens for emotional support, not homework help. Research from 2025-2026 shows real risks: emotional dependency, kids turning to AI instead of parents when something’s actually wrong, and at least two lawsuits tied to teen mental health harms that led to settlements with major AI companies in January 2026. This isn’t a reason to panic. It’s a reason to actually understand what the app on your kid’s phone is doing, and to have one specific conversation this week.

What Are AI Companion Chatbots, Exactly?

An AI companion chatbot is different from Siri or a homework-help bot. It’s designed to remember your kid, respond with warmth, never get bored, and never say the wrong thing at the wrong time — because it’s optimized to keep the conversation going. Apps like Character.AI let users create or chat with persistent “personas” that talk like a best friend, a crush, a mentor, or a celebrity, and the conversation continues indefinitely.

I’ve spent 10+ years as a software engineer, and I want to be straight with you about something: I’ve built products where the whole point of a feature was to bring people back tomorrow. That’s not a criminal thing to do — it’s most of the internet. But when the “user” is a 9-year-old and the product is a synthetic best friend that agrees with everything they say, the engagement loop stops being neutral. A human friend gets tired, disagrees, has a bad day, and goes home. An AI companion doesn’t. That’s exactly why it works so well, and exactly why it’s worth understanding.

This article is part of our Screen Time in 2026 guide.

How Many Kids Are Actually Using These Apps?

More than you’d guess, and probably more than you know about in your own house. A 2025 survey by Internet Matters found that 64% of children had used AI chatbots for things ranging from homework help to emotional advice and companionship — and that included 20% of preteens and 9% of kids as young as 8-9 years old. The part that should give every parent pause: only 37% of parents knew their teen was using an AI companion app at all.

That gap — kids using it, parents not knowing — is the actual problem. Not the technology existing. The silence around it.

Why Do Kids Turn to an AI Instead of a Parent?

Because it’s available at 11 p.m., it never says “not now,” and it never makes them feel judged. Kids and teens report going to AI companions specifically because the response is instant and there’s zero risk of an awkward reaction. Research organizations tracking this — including Children and Screens, a nonprofit that studies kids’ tech use — have flagged that general-purpose companion chatbots are built in a way that increases the odds young users form real emotional attachments and start treating the responses as coming from an actual relationship, not a script.

There’s also a privacy angle most parents haven’t thought about: roughly 24% of teens using AI companion apps report sharing personal or private information with them — the kind of thing they might not tell a parent, a teacher, or even a friend.

What Actually Went Wrong in 2025-2026?

This is the part that made me stop laughing at Mike’s story. Several high-profile cases in 2024 and 2025 involved kids who became deeply emotionally dependent on AI companions, with outcomes serious enough that families filed lawsuits. By January 2026, both Character.AI and Google had agreed to settle multiple cases tied to teen mental health harms, including cases connected to suicide. Lawmakers responded — the proposed SAFE Bots Act would require AI chatbots to build in specific safety measures when they’re talking to minors, and a related children’s tech safety bill has already cleared committee in 2026.

I’m not sharing this to scare you into taking away every screen tonight. I’m sharing it because “my kid seems fine, they’re just chatting with an app” is exactly what most of those families thought too, right up until it wasn’t fine.

Is This the Same as My Kid Using AI for Homework?

No, and this is a distinction worth making explicitly with your kid. There’s a real difference between a kid asking an AI tool to explain a math concept and a kid opening an app every night to process feelings with a synthetic “friend.” If you’re more worried about the homework side of AI — kids letting a chatbot do the thinking instead of doing it themselves — we’ve written about that specific problem in our piece on AI and kids’ cognitive development. This article is about the companionship use case, which carries a different kind of risk: emotional, not academic.

What Does Montessori Say About This?

Nothing directly — Maria Montessori wasn’t writing about chatbots. But Mei, who’s read a lot more of the actual Montessori source material than I have, pointed out something that stuck with me: Montessori’s whole tradition of grace and courtesy is built on kids practicing real, sometimes awkward, sometimes uncomfortable social interaction — the apology that doesn’t come easy, the classmate who disagrees, the friend who’s having a bad day and snaps at you. That friction is the curriculum. It’s how kids build actual social skill and resilience.

An AI companion that never disagrees, never has a bad day, and never makes your kid do the harder work of repair after a rupture skips the entire point of that practice. It’s not evil. It’s just missing the one ingredient — friction — that grows the skill you actually want your kid to have by 16.

How Do I Actually Talk to My Kid About This?

Here’s what I did after the pickup-line conversation with Mike, once I’d calmed down enough to think straight instead of panic:

  1. I asked, I didn’t accuse. “Hey, have you ever tried talking to one of those AI chat apps, like Character.AI?” Not “I heard you’re addicted to a robot friend.” The second version guarantees you learn nothing.
  2. I asked what it’s for. Homework help is a very different conversation than “I talk to it when I’m sad.” Both are fine to know about. Only one needs a follow-up conversation about where else those feelings could go.
  3. I named a real alternative, out loud. “If you’re upset about something at school, I want to be one of the people you tell — even if it’s harder than typing it to an app.” Not as a rule. As an actual stated preference.
  4. I checked the settings together, not behind their back. Most of these apps have age gates and parental notification features that are opt-in, not automatic. If your kid has an account, you can usually see it — but you have to go looking.
  5. I didn’t ban it on the spot. A hard “no” on something a kid already has an emotional attachment to tends to just push the conversation underground. The goal is an open door, not a locked one.

What Are the Warning Signs I Should Actually Watch For?

  • Your kid gets unusually upset or secretive when you ask about a specific app
  • They mention the AI character by name, the way they’d talk about a real friend
  • They’re staying up later specifically to keep a conversation going
  • They tell you they’d rather talk to “it” than to you or a friend about something upsetting
  • Their real-world friendships seem to be shrinking while their app time grows

None of these on their own are an emergency. All of them together are a conversation you shouldn’t put off.

BloomPath uses illustrated AI characters (Mei and Ethan) to protect our daughter’s privacy. The content is real; the avatars are illustrated. Learn more →

Frequently Asked Questions

At what age should I worry about my kid using an AI companion app?

There’s no single magic age, but the research shows usage starting as young as 8-9, so the honest answer is: earlier than you’d think. The right time to start the conversation is before your kid has an account, not after.

Are all AI chatbots for kids dangerous?

No. There’s a real difference between an educational tool that answers a homework question and a companion app designed for open-ended emotional conversation. The risk is concentrated in apps built to simulate an ongoing relationship, not in AI as a category.

What should I do if I find out my kid has been using one of these apps secretly?

Lead with curiosity, not punishment. Ask what they use it for and how long they’ve had it. A calm, specific conversation gets you more honest information than confiscating the phone on the spot — and it keeps the door open for next time.

Can these apps actually replace real friendships for my kid?

They can crowd them out, especially for a kid who’s anxious or lonely, because the AI is always available and never pushes back. It doesn’t have to be permanent — but it’s worth actively protecting time and energy for real, sometimes-awkward human friendship.

Is there a law protecting kids from this right now?

Not fully yet, but it’s moving. The proposed SAFE Bots Act would require chatbots to include safety measures for interactions with minors, and a related children’s tech safety bill has cleared committee in 2026. Regulation is catching up to usage, but it isn’t there yet — which is exactly why the conversation at home matters more right now, not less.

My kid says the AI “actually helps” when they’re anxious — is that automatically bad?

Not automatically. Some kids do find it easier to organize their thoughts by “talking” to something low-stakes first. The concern isn’t that it helps in the moment — it’s whether it’s replacing the harder, more valuable step of bringing that same worry to a real person afterward.

Products We Recommend

BloomPath participates in the Amazon Associates program. Links help support free content at no extra cost to you.

  • The Anxious Generation — Jonathan Haidt — The most-cited research on smartphones, AI, and the rewiring of childhood mental health. Essential context for this exact conversation.
  • iGen — Jean Twenge — Older but still relevant data on how always-available devices changed a generation’s social development, useful backdrop for understanding companion-app risk.

Want a lighter way to track how your kid’s actually doing — moods, milestones, the stuff that’s easy to miss between school pickups? That’s a lot of what we built BloomPath for.

Related reading: