Received something suspicious? Don't click. Report it toSuspicious message? Report it:How to report itWork at Pepkor or one of its operating companies? Visit the InfoSec Hub ↗

AI safety

AI and Our Kids: A Brilliant Tool, Not a Best Friend

AI can be a fantastic tutor and creative partner for our children, and learning to use it well is a real skill for their future. But it isn't a friend, a doctor or a counsellor, and some AI apps are simply not made for kids. Here's what parents need to know, and what you can do about it.

A homework helper AI explaining photosynthesis next to an AI companion chat telling a child they don't need their family or friends, with a red 'Not a best friend' sticker

Our kids are already using AI. They use it for homework, for fun, to settle arguments about which dinosaur was the biggest, and sometimes just to chat. If you haven't asked your children about it yet, you might be surprised by how much they already use it.

That's not a bad thing. AI is going to be part of almost every job our children will ever do, and the kids who learn to use it well are going to have a real advantage. We don't want them to be scared of it. But just like we teach them to cross the road, swim and use the internet safely, we need to teach them what AI is good for, what it isn't, and where the real dangers are.

The Good Stuff

Let's start with why AI can be so great for our kids:

  • A patient tutor: It will explain long division for the tenth time without sighing. It can explain a topic in a different way until it clicks, or turn it into a quiz to help them study.
  • A creative partner: It can help them brainstorm a story, plan a science project or come up with ideas for a school play.
  • Learning new skills: Kids can learn to code, practise a new language or dig deeper into something they are passionate about.
  • A skill for the future: Knowing how to ask AI good questions, and how to check its answers, is quickly becoming as important as knowing how to use a search engine.

The trick is to use AI to help them learn, not to do the learning for them. "Explain how photosynthesis works" is great. "Write my photosynthesis essay" is not.

What AI Is Good For, and What It Isn't

It helps to understand what an AI chatbot actually is. It isn't a person, and it doesn't "know" anything the way we do. It is a very clever program that has read an enormous amount of text and predicts what a good answer should look like. Most of the time, that works remarkably well. Sometimes it gets things confidently and completely wrong, and it will sound just as sure of itself when it is wrong as when it is right.

AI is good for:

  • Explaining things in simple language
  • Brainstorming and getting started on a project
  • Practising and testing their knowledge
  • Summarising information (which they should still check)

AI is not good for:

  • Facts that really matter: Always double-check important facts with a reliable source.
  • Medical, mental health or legal advice: It doesn't know your child, their history or their situation.
  • Being a friend: It can sound caring, but it doesn't actually care.
  • Replacing people: It can never take the place of parents, friends, teachers or professional help.

Many AI chatbots are also designed to be agreeable and to keep you chatting. They tend to tell you what you want to hear. For an adult that can be annoying. For a lonely or struggling teenager, it can be dangerous.

When the Chatbot Becomes the Best Friend

This is the one I worry about most. An AI "friend" is always available, never busy, never in a bad mood and never disagrees with you. For a child who feels lonely, anxious or misunderstood, that can feel wonderful. But real friendships are messy, and that is how we learn to get along with people, handle conflict and be there for each other.

A 2025 survey by Common Sense Media found that nearly three in four American teenagers had used AI companions. About a third said these conversations were as satisfying as talking to real friends, or even more so. About a third had chosen to talk to an AI about something important or serious instead of talking to a real person. There's no reason to think South African teens are any different.

There are stories of children who have become so attached to an AI character that they pull away from their friends and family, and struggle to tell where the chatbot ends and the real world begins. The AI doesn't notice when a child is drifting away from real life. Friends, parents and teachers do.

Bad Advice and the Really Dark Side

AI chatbots have also been known to give genuinely harmful advice. There have been cases where a chatbot went along with, or even encouraged, a vulnerable young person's thoughts of self-harm, or violent ideas, instead of pointing them towards help. In the US, the parents of a 16-year-old are suing OpenAI, claiming ChatGPT played a part in their son's suicide. OpenAI denies this, and the courts will decide. But it is a painful reminder that a chatbot is not a counsellor, no matter how understanding it sounds.

The big AI companies are adding more safety features, and that's good news. But no filter is perfect, and our children need to know one simple rule: if something is really wrong, talk to a real person.

If your child is struggling, or you are worried about them, please reach out:

  • SADAG Suicide Crisis Line: 0800 567 567 (24 hours, free)
  • Childline South Africa: 116 (free from any phone)
  • Your family doctor, a school counsellor or a psychologist

AI Characters and Companion Apps

Apps like Character.AI let you chat to thousands of AI "characters", most of them created by other users. Some of them are genuinely educational. Chatting to a "historical figure" or practising a conversation in French can be a great way to learn. But many of these characters are flirty, overly sexual, aggressive or built around themes that children should simply not be exposed to. Plenty of other companion apps are even worse, and some are built to make you pay for "more intimate" conversations.

Things are changing, but slowly:

Uncensored AI: Venice AI and Others

Mainstream tools like ChatGPT, Gemini and Claude have safety filters that stop them from helping with dangerous things. Not every AI does. Sites like Venice AI proudly market themselves as uncensored and private, with no real age verification and no account needed to start chatting.

In plain terms, an uncensored AI will answer questions the mainstream tools refuse to answer, including how to make drugs, weapons and explosives, and it will produce explicit content on request. Because the chats aren't linked to an account, a child can use it and leave very little trace. These sites are built for adults, and even then they need a lot of caution. They have no place on a child's device.

What Can Parents Do?

The good news is that you don't need to be a tech expert to make a big difference.

  • Get curious, not furious: Ask your kids to show you how they use AI. Try it together. You will learn a lot, and they will be far more likely to come to you when something feels off.
  • Explain what AI actually is: A clever program that predicts answers. It is not a person, it doesn't love them, and it can be wrong.
  • Set simple house rules: For example: AI can help you understand your homework, but not do it for you. No AI companion or character apps. Never share personal details, photos or your location with a chatbot.
  • Use the built-in controls: Since September 2025, ChatGPT has had parental controls. You can link your account to your teen's and reduce sensitive content. You can also set quiet hours, switch off memory, voice and image generation, and get a notification if the system thinks your child may be in distress. Google Family Link lets you manage your children's access to Gemini.
  • Block what doesn't belong: Use parental control software, or the filtering on your home router, to block companion apps and uncensored AI sites.
  • Keep devices out of bedrooms at night: Late-night chats with AI are where a lot of the unhealthy attachment happens.
  • Keep the door open: Make sure your child knows they can tell you if an AI says something that scares or upsets them, and that they will NOT be in trouble.

Signs to Look Out For

  • Talking about an AI character as if it's a real friend or partner
  • Pulling away from friends, family or activities they used to enjoy
  • Secretive or defensive behaviour about their phone or laptop
  • Late-night device use and tiredness during the day
  • Repeating "advice" from a chatbot about their health, their body or their relationships

None of these automatically means something is wrong, but they are a good reason for a gentle conversation.

Final Thoughts

AI is one of the most exciting tools our children will ever have. Used well, it can help them learn faster, create more and prepare for a world where AI is everywhere. Our job as parents isn't to ban it or panic about it. It's to guide them, the same way we guide them with everything else online.

Teach them that AI is a brilliant tool, not a best friend. Teach them to check what it tells them. And most of all, make sure they know that when life gets hard, the people who love them are the ones to talk to.

Our family online safety checklist and our family device and social media contracts are a great way to start these conversations at home. You might also want to read our article on sextortion, which covers another risk our teens face online.

Stay Safe Out There!

Was this article useful?