ChatGPT age limit: what the rules say, and what to do if your child is under 13

Short answer: the ChatGPT age limit is 13, and OpenAI requires a parent's consent for anyone aged 13 to 17. Under 13, the honest reading of the rules is that your child does not get an account, and if they use ChatGPT at all it is alongside you, on your account. That is also what I do at home, and not only because the rules say so.

I have written a longer post on whether ChatGPT is safe for kids at each age. This one is narrower. It is about the age rule itself: what it actually protects against, what it cannot, and what to do if your child is on the wrong side of it.

What the rules say

I checked OpenAI's own pages on the date at the top of this post. They change, so check them again before you rely on them.

Under 13. ChatGPT is not meant for children under 13. There is no children's version and no supervised account for a nine-year-old. If a teacher uses it with a class of under-13s, OpenAI says the adult must be the one operating it. That is the same arrangement I recommend at home.

13 to 17. A teenager may have an account, but OpenAI requires a parent's consent. Since August 2026 the company has been rolling out ChatGPT for Teens, which switches on by itself when the account's stated age, a verified age, or OpenAI's own age-prediction system says the user is under 18. It adds safeguards around sensitive topics such as self-harm and eating disorders, plus study tools that give hints rather than answers; a wrongly flagged adult can verify their age.

Parental controls. You can link your account to your teenager's from Settings and switch on things like study mode and quiet hours. You cannot read their chats. The details are in the safety post, so I will not repeat them here.

One thing worth knowing about the number itself: 13 is not a developmental finding. As far as I can tell it is largely a legal line, because children's data-protection rules in many countries apply below that age, so companies draw the line there. It is a gate, not a verdict on your child.

What the age limit protects against, and what it cannot

The rule does three useful things. It keeps the youngest children from opening accounts in their own name. It triggers the teen safeguards, which are real, and with age prediction now in the mix a teenager who fudged a birthday is more likely to be caught than they were two years ago. And it gives you, the parent, something to point at: "The rule is 13. That is not me being difficult."

Now the honest part. An age limit is a box a child ticks. A determined ten-year-old can use a parent's logged-in laptop, an older sibling's phone, or a friend's account at a sleepover. The rule cannot see any of that.

More importantly, the age limit says nothing about how the tool is used. It cannot tell the difference between a ten-year-old making up a story with her mum and a ten-year-old pasting a maths worksheet in and copying the answers out. The three risks that actually matter with ChatGPT are quiet ones: it does the thinking for a child, it is confidently wrong, and it is a stranger the child tells things to. None of those switches off at 13. Some of them get worse, because at 13 the supervision drops away.

Under 13: what to do instead

Three things, and they build on each other.

Use it together, on your account. Do not hide the tool from a curious child; that just makes it interesting. Sit at the kitchen table with your own account open and ask it for a story about a dragon who is scared of toast. Then ask it something you know it will get wrong, a fact about your town, a date from your own family, and watch your child's face when it answers with total confidence and total wrongness. That moment is the whole lesson.

The concept you are planting is the one the book returns to again and again: this is a clever machine that makes things up. It is not a person. It is not always right. Chapter 16 offers the metaphor I use with the youngest children: a very well-read robot that has read millions of books but has never been to a park, so it does not know what grass smells like.

Ages 5 to 8 get the toy version only: silly pictures, stories. Ten minutes, together, then off. The "What Does It Not Know?" game from Chapter 16 fits here. Ask the AI the capital of France, then ask it what it feels like to be really nervous, and compare the answers. Ask your child why it could manage the first and not really the second.

Ages 9 to 12 are ready for the supervised apprentice stage from Chapter 10, still on your account or beside you. This is where the 20-Minute Rule starts: before your child asks an AI anything about a piece of schoolwork, they spend twenty minutes on it alone, and then they ask for help understanding, not for the answer. Shared device, shared room, and you nearby and curious rather than hovering.

Use children's tools where they exist. Chapter 18 makes the point plainly: tools designed for minors, with age-appropriate filtering and data protections, are meaningfully safer than a general-purpose chatbot used by a child. ChatGPT has no such version. Google's Gemini, at the time of writing, does let under-13s use it through a supervised Family Link account that a parent must switch on first; I have a separate post on whether Gemini is safe for kids.

Why the rule at home matters more than the rule on the website

Here is the thing I most want you to take from this post. OpenAI's rule is about whether your child may use ChatGPT. The rule that protects your child is about how, and only you can set that one.

Chapter 10 puts it in a single sentence I say out loud at home more than any other: "Is this helping you think, or thinking for you?" A ten-year-old who can answer that question honestly is safer with ChatGPT than a fifteen-year-old who has never been asked it.

The home rules travel with your child past the age gate, into the years when nobody is checking.

The 20-Minute Rule, above, so the struggle happens before the shortcut is available.

The V.E.T. rule for anything an AI produces: Verify it against a second source, Explain it in your own words, Transform it into your own work. If your child cannot explain it, they have not learned it. There is a whole post on teaching the V.E.T. rule.

The Family AI Agreement, which Chapter 10 gives as a one-page template and which you can build for free at /tools/family-agreement: which tools, for what, what never gets typed in, and the "Uh-Oh" plan for when something goes wrong. With an under-13 it will be short. Write it anyway. The conversation you have while filling it in is the point.

And every rule comes with a why, because a rule with a reason survives when you are not in the room. Not "because the website says 13". Rather: "because we value the satisfaction of solving hard things ourselves, and our brains get stronger when they struggle."

Questions parents ask

How old do you have to be to use ChatGPT?

At the time of writing, OpenAI says ChatGPT is not meant for children under 13, and users aged 13 to 17 need a parent's consent. Accounts identified as under 18, by stated age or by OpenAI's age prediction, get the ChatGPT for Teens experience automatically. The company's pages change, so check the current version before relying on it.

Can a 13 year old use ChatGPT?

Yes, with a parent's consent, and the account will be switched to the teen version with extra safeguards and study tools. I would link your account to theirs from Settings the same week, turn on study mode and quiet hours, and agree the 20-Minute Rule and the V.E.T. rule out loud. The account is allowed at 13; the judgement still needs building.

Can my 10 year old use ChatGPT on my account?

That is the arrangement OpenAI describes for classrooms, an adult operating the tool, and it is what I do at home: together, on my account, in a shared room, for short creative sessions or for help understanding schoolwork after twenty minutes of trying alone. Not alone, not for finished answers, and not with anything personal typed in.

What happens if my child lied about their age on ChatGPT?

OpenAI's age-prediction system may flag the account as under 18 and switch on the teen experience anyway, which is a good outcome. What matters more is the conversation: not punishment, but "what have you been using it for?" Then look at the chats together and decide, as a family, what the rules are from here.

Sources

• Chapter 10 of Raising AI-Ready Children on the Digital Guardian's role, the 20-Minute Rule, the V.E.T. checklist, age-appropriate guidelines and the Family AI Agreement.
• Chapter 16 on the developmental picture from ages 5 to 18 and the "well-read robot" metaphor for young children.
• Chapter 18 on using children's versions of tools where they exist.
• Ying Xu, Harvard Graduate School of Education, research on children's interactions with conversational AI, as cited in Chapter 16.
• OpenAI help centre, "Is ChatGPT safe for all ages?", "ChatGPT for Teens" and "Managing parental controls in ChatGPT", checked on the date at the top of this post.