Skip to content

Guide for parents

Is AI Safe for Kids? A Parent's Guide to the Real Risks

Updated

By . Founder of ThinkaBit: a dad based in the Netherlands who has worked in tech for 25 years and now works with AI.

“Is AI safe for kids?” is a reasonable question with an unsatisfying honest answer: it depends on which AI, how it’s used, and how much the child understands about it. AI isn’t one thing. It’s the recommendation system choosing the next video, the filter on a photo app, the chatbot a classmate uses for homework, the “character” in a companion app, and the voice clone in a scam call. Each comes with different risks.

This guide sums up what we’ve written for parents of children aged 7–12 and links to the posts that go into each topic. The short version: most of the risks are manageable, and the most effective protection isn’t a ban or a setting, but a child who understands what AI is and knows when to come to you.

The real risks, roughly in order of how often kids meet them

1. Content chosen by AI. Long before a child types anything into a chatbot, AI is deciding what they see: the next video, the suggested game, the “you might like” row. These systems are designed to keep people watching, not to show what’s best for a child. The risk isn’t a single dramatic moment; it’s hours of passive, endless scrolling and content that slowly drifts away from what you’d pick.

2. Fakes and misinformation. AI makes it cheap to produce convincing pictures, videos and voices. Children can be fooled by fake news clips, fake “facts” and fake people. Our guide on how to teach your kids to spot deepfakes and AI fakes covers this in detail.

3. Privacy and data. Many apps, including some made for children, collect more data than they need: location, contacts, voice recordings, photos, detailed behaviour. Some of it is used to build profiles for advertising or to train AI systems. Children rarely understand what they’re giving away.

4. Chatbots that sound sure but get things wrong. General-purpose chatbots like ChatGPT can be useful, but they make things up with complete confidence, can give advice that isn’t suitable for a child, and weren’t designed for young users.

5. Companion chatbots. Apps that let children chat with “characters” or “friends” are a different category: they’re built to feel personal and to keep the conversation going, which makes them riskier for children than a question-answering tool.

6. Scams. Voice cloning and fake messages make scams more convincing, and children are targeted through games and messaging apps.

Is AI safe for kids? A parent’s guide goes through each risk and how big it really is, and sets out a five-step plan.

Chatbots: ChatGPT and companion apps

Most general-purpose chatbots, ChatGPT included, set a minimum age of 13, with parental permission for teenagers. They have no reliable way of knowing a child is typing. That doesn’t mean a 9-year-old must know nothing about chatbots, but it does mean any use should happen alongside a parent, on a parent’s account, with clear ground rules: never share personal information, double-check anything important, and treat the answers as a first guess rather than the truth.

Should kids use ChatGPT? explains the age rules, the real risks and six ground rules, and walks you through a first session to try together, so your child’s first experience of a chatbot is one where you’re both thinking about what it says.

Companion apps need extra care. A chatbot that remembers your child, says it missed them and is always available can become something a child confides in instead of the people around them. Is Character.AI safe for kids? covers what these apps are, what their age rules say, why they’re riskier, the signs that a child’s use has become more than a game, and what to do if your child is already using one. Rules on these apps change quickly, so check each app’s own pages too.

Privacy: what to check before you install

You don’t need to read every privacy policy in full to make a good decision. A few questions catch most problems: Does the app show ads? Does it ask for permissions it doesn’t need (location, contacts, microphone)? Can you see and delete your child’s data? Does it say clearly what it collects and why? Is it built to meet children’s privacy laws such as COPPA in the US and GDPR in Europe?

AI and kids’ privacy: what data are apps collecting? explains what apps can collect, what a kid-safe app should do, what the law requires in short, the red flags to look for and a two-minute check to do before installing anything.

It’s also worth talking to your child about privacy in their own terms: that a nickname is safer than a real name, that a photo can show where you live, and that “free” apps often pay for themselves with data.

Scams that use AI

AI has changed what scams look like more than how they work. The tricks are old: urgency, secrecy, a prize, a request for personal information. What’s new is that a message can now sound exactly like a friend, a voice can sound like a relative, and a video call can show a face that isn’t there.

The rules that protect children still work: check on a second channel (call the person back on a number you know), treat urgency as a warning sign, never keep online secrets from parents, never give out personal information, and be suspicious of anything “free” or “you’ve won”. A family code word adds an easy extra layer. Teaching kids online safety in the age of AI scams explains each rule with examples children understand.

Screen time: quality matters more than minutes

Worrying about AI often turns into worrying about screen time. Limits help, but the minutes matter less than what happens in them. An hour of a child creating, solving problems or learning something with you is very different from an hour of autoplay chosen by an algorithm.

Screen time vs. screen quality explains the difference between active and passive screen use, gives five questions to judge any app or game, explains why “endless” designs are a problem, and suggests a simple family screen plan.

A calm plan for parents

Pulled together, the advice across these posts comes down to five habits:

  1. Talk early and often. Short, regular conversations work better than one big talk. Ask what they’ve seen, not just what they’ve done.
  2. Set a few clear rules. No personal information online, no secrets from parents about online contact, and ask before installing anything new.
  3. Teach understanding, not just avoidance. A child who knows that AI guesses from patterns is much harder to fool than one who’s only been told “don’t”. Our guide on how to explain AI to a child shows how.
  4. Choose kid-safe tools. Prefer apps without ads or tracking, that are built for children’s privacy laws, and that put you in control.
  5. Keep your child in the driver’s seat. The aim is a child who uses technology with confidence and judgement, not one who’s scared of it.

How ThinkaBit fits in

ThinkaBit is our app for kids aged 7–12, built around that third habit: understanding. Children learn how algorithms make decisions, what data apps collect, why AI can be biased or wrong, how to stay safe online and how to spot fakes, through short games with Sparky, Glitch and Professor Pixel. ThinkaBit makes no generative-AI calls, children never chat with an AI in the app, and there are no ads and no tracking. It’s built to meet COPPA (US) and GDPR (EU), and the Grown-ups area sits behind a PIN.

For help beyond this site, our resources page lists trusted organisations and helplines for parents in several countries.

Practise it with ThinkaBit

Short, playful lessons for kids aged 7–12 on how AI works, spotting fakes and staying safe online. No ads, no tracking, free to start.

Share: