The Right Way for Children to Use AI: An Age-by-Age Guide for Parents

Here's something that surprises many parents. AI can make your child worse at a subject. It can also help your child practise more without doing any harm. What makes the difference? Not the app. The way your child uses it.

This guide shows you how, in plain steps, age by age.

One rule that explains everything

In our last article, we told you about a big study in Turkey. Nearly 1,000 secondary school students practised maths. Some had a plain ChatGPT helper. Some had an AI tutor that was built to give hints, not answers. Some had no AI.

Here's what happened:

• Plain ChatGPT: practice marks went up 48%. But on the test without AI, they scored 17% lower than students with no AI.
• Hint tutor: practice marks went up 127%. And on the test, they did about as well as students with no AI.

By our sum, the hint tutor gave students more than two and a half times the practice boost of plain ChatGPT, without the drop on test day.

So the whole guide comes down to one rule:

Your child does the thinking. AI only helps.

When AI gives answers, your child stops thinking. When AI gives hints, your child keeps thinking. Hold on to that, and everything below makes sense.

How many children already use it?

More than you might think. In America, the Pew Research Center found in 2026 that more than half of teenagers (54%) use AI chatbots for schoolwork. About 1 in 10 do all or most of their schoolwork with them.

By our sum, that means about 1 in 5 of the teens who use AI for school are letting it do most of the work.

We don't have the same survey for Nigerian children yet. But we know 88% of Nigerian adults have used an AI chatbot, far above the world average. Many children use their parents' phones. So it's safest to assume your child has already met AI, or soon will.

What the apps' own rules say

Before anything else, know the age rules. These come from the companies themselves:

• ChatGPT: 13 and older. Anyone under 18 needs a parent's or guardian's permission.
• Meta AI (from the company behind WhatsApp, Facebook and Instagram): 13 and older.
• Gemini (Google): children under 13 can only use it if a parent turns it on through Google's Family Link. Google doesn't say whether this is offered in Nigeria, so check before you count on it.
• Claude: 18 and older. It isn't meant for children or teenagers.

Here's a sum that shows why this matters. DataReportal's 2026 figures show children aged 5 to 12 make up 21.2% of Nigeria's people. By our count, that's about 51 million Nigerian children who are too young for ChatGPT or Meta AI under the companies' own rules.

Nigeria's own law agrees that children need protecting. Under the Nigeria Data Protection Act 2023, anyone under 18 is a child, and a company needs a parent's or guardian's consent to use a child's personal data.

The age-by-age guide

Experts don't all use the same age groups, but they agree on the big picture. Here it is, based on Common Sense Media, the American Academy of Pediatrics, and researchers at the Children's Hospital of Philadelphia (CHOP).

Age 5 and under: no chatbots

Common Sense Media keeps it short: "No AI chatbots". The CHOP researchers say that at this age, "human interaction should be prioritized."

In plain words: talk to your child, read with them, play with them, let them ask you a hundred questions. That's how small children learn to think and talk. A chatbot can't replace you.

Ages 6 to 12: only with you beside them

Common Sense Media says: "Limited use, alongside a parent or caregiver."

The CHOP researchers explain why. They warn that children in this age group "may have difficulty identifying misinformation generated by AI." In plain words, AI can say something wrong in a very confident voice, and a young child can't tell.

So at this age:

• Sit with your child when they use AI. Make it something you do together.
• Use it to explore a question, then check the answer in their textbook.
• Ask, "Do you think that's true? How can we check?"

Remember, ChatGPT and Meta AI don't allow under-13s at all. If your child is under 13, stick to tools built for children and use them together.

Ages 13 to 17: homework help, with rules

Common Sense Media says: "Supervised use is OK for homework and creativity, not for comfort or feelings."

At this age, your teenager can start using AI for school, with your permission and clear rules. More on the rules below.

Every age: never for feelings

This is the firmest line in all the expert advice. Common Sense Media says AI chatbots should be:

"Never for companionship, emotional support, or mental health advice." Common Sense Media, 2026

Why so strict? In our last article, we shared a 2025 study that tested 29 chatbots on how they'd answer someone thinking about suicide. Under strict rules, none of them gave a good enough answer. And a 2026 US survey found that most young people who ask chatbots about their feelings tell no adult.

Tell your child plainly: "When you're sad, worried or scared, talk to a person. Talk to me."

The "think first" rule for homework

This is the habit that protects your child's thinking. Teach it, and check it.

1. Try first. Your child attempts the question on paper, alone. Even a wrong attempt counts.
2. Ask for a hint, not the answer. If they're stuck, they ask AI for one small hint.
3. Check. They compare the AI's help with their textbook, notes or teacher. AI can be wrong.
4. Explain it back. At the end, they explain the answer to you in their own words. If they can't, they haven't learned it yet.

Here's a message your child can type at the start of a chat, so the AI acts more like that hint tutor:

"Please don't give me the answer. Give me one small hint at a time, and let me try first."

This isn't the same tool the researchers built. But it follows the same idea: the AI guides, your child does the work.

Good uses and bad uses

Good ways to use AI:

• "Explain photosynthesis in simpler words." (after reading the textbook)
• "Give me five practice questions on fractions."
• "Check the grammar in this paragraph I wrote."
• "Give me a hint for step 2 of this maths question."

Bad ways to use AI:

• "Write my essay on my best holiday."
• "Solve my homework questions."
• "Summarise this chapter so I don't have to read it."
• Talking to a chatbot instead of a person when feeling low.

See the pattern? Good uses keep your child doing the thinking. Bad uses hand the thinking to the machine.

ChatGPT's parental controls

If your teenager uses ChatGPT, it has parental controls. OpenAI launched them in September 2025. According to news reports:

• You and your teen link accounts. One of you sends an invite and the other accepts.
• You can reduce graphic content, dangerous viral challenges, and sexual, romantic or violent role-play.
• You can set "quiet hours" when ChatGPT can't be used, like bedtime or school time.
• You can turn off voice chat, memory and making pictures.
• If your teen unlinks the accounts, you get told.

Controls help, but they're not a babysitter. Talking with your child still matters most.

Do you need to pay for AI?

No. Free versions are enough for homework help.

On Nigeria's App Store, ChatGPT Go costs ₦6,900 a month and ChatGPT Plus costs ₦31,500 a month (checked 3 November 2026).

By our sum, Plus costs 45% of a month's ₦70,000 minimum wage. Over a year, that's ₦378,000, more than five months of minimum wage. Go costs about 10% of the minimum wage each month.

Paying more doesn't make AI safer for your child. Good habits do, and they're free.

"But I don't use AI myself"

That's okay. Many parents don't. GSMA, the group that represents mobile networks, reported in 2026 that about 140 million Nigerians lived in places with mobile internet but didn't use it. And in its survey of adults, 66% in towns used mobile internet, against 33% in villages. By our sum, a town adult is twice as likely to be online as a village adult.

So if you're not online much, you're not alone, and you can still guide your child. You don't need to understand how AI works to ask, "Show me what you asked it," or, "Explain this to me in your own words."

The American Academy of Pediatrics has a simple tip for parents: "Try AI together." Ask your child to teach you. Children love being the teacher, and you'll see exactly how they use it.

What Nigeria's government says

Nigeria's education ministry has said AI should support teachers, not replace them. In January 2025, a statement read for Education Minister Tunji Alausa said:

"AI enhances, rather than replaces, educators' vital role in shaping young minds." Dr Tunji Alausa, Minister of Education, January 2025

In July 2026, the government said it would train 11,700 Unity College teachers in AI. But Nigeria's draft national AI strategy says little about schoolchildren, according to a 2025 study of it. So for now, the most important rules are the ones you set at home.

AI rules for kids at home: a simple plan

1. Know the age rules. No chatbots for under-6s, only with you for under-13s, and your permission for teens.
2. Teach "think first". Try alone, then a hint, then check, then explain it back.
3. Never for feelings. Feelings go to people, not machines.
4. Make it normal to say "I used AI". A child who can tell you is a child you can guide. A child who hides it is one you can't see.
5. Talk about checking, again and again. AI can be wrong. One talk isn't enough.

At Scholarleveling, children learn in live classes with real teachers, who can see whether a child understands, not just whether the answer is right. Parents also get a code to see their child's progress every week.

Read next: what AI is doing to children's thinking, or how coding helps a child's brain.