# Can Kids Use ChatGPT? The AI Age Limits, and What They Actually Do

**Author:** Dan Fitzpatrick
**Canonical URL:** https://parents.theaieducator.io/posts/can-kids-use-chatgpt
**Published:** 2026-09-15T07:25:49.000Z
**Updated:** 2026-09-15T07:25:58.397Z
**Category:** Safety and Wellbeing

## TL;DR

Most AI tools set 13 as the minimum. ChatGPT asks for a parent's permission up to 18, Claude is adults only at 18, Gemini allows under-13s on a supervised account that a parent has to switch on, and Character.AI has removed open-ended chat for under-18s. None of these is checked at sign-up, so the age limit mainly decides which version of the product your child gets rather than whether they get in. Do this first: open the AI app your child uses and look at the date of birth actually stored on the account, because every filter and teen safeguard hangs off that one number. If your child also has a school account, that is a second age setting, and the school chose it. Where you are worried about your child rather than about a setting, school rules and professional advice come first.

## Key takeaways

- The age limits are not the same: ChatGPT is 13 with parental consent to 18, Claude is 18, Gemini allows under-13s on a parent-enabled supervised account, and Character.AI has ended open-ended chat for under-18s.
- Almost all of these are self-declared at sign-up, so an age limit is a promise, not a lock.
- Run the Stated Age Check tonight: look at the date of birth actually stored on your child's account, not the one you assume.
- On a school-issued account the age category is set by a school administrator, not by you, and it can differ from the consumer age limit.
- Thirteen comes from a US privacy law, not from a judgment that 13 year olds are ready.

## Article

Yes, most children can use ChatGPT, but the honest answer is that "can" is doing a lot of work in that sentence. OpenAI's own rule is that ChatGPT is not meant for children under 13, and that 13 to 18 year olds need a parent's permission. Nothing checks either one at sign-up. Your child types a date of birth, and that date decides what the app does next.

## What are the actual age limits for the main AI tools?

They are all different, and three of the big ones changed in the last year. These are the companies' own published rules, read on September 15, 2026.

**ChatGPT (OpenAI): 13, with a parent's permission up to 18.** [OpenAI's help center](https://help.openai.com/en/articles/8313401-is-chatgpt-safe-for-all-ages) states plainly that "ChatGPT is not meant for children under 13," and asks that "children ages 13 to 18 obtain parental consent before using ChatGPT." It also says, in the same place, that "ChatGPT may produce output that is not appropriate for all audiences or all ages."

**Gemini (Google): under 13 is allowed, but only if you switch it on.** This one surprises most parents. A child under 13 can use Gemini Apps on a supervised Google Account, and a parent has to enable it through Family Link first. [Google's own guidance for families](https://support.google.com/families/answer/16109150?hl=en) is unusually blunt about what that does not buy you: "Our filters try to limit inappropriate content from appearing, but they're not perfect. Your child may encounter content you don't want them to see." It also tells parents to tell their child that "Gemini should not be treated as a friend or used to discuss personal problems or secrets." One regional catch, as of September 2026: Gemini Apps are not available for supervised accounts in the European Economic Area, Switzerland or the United Kingdom, so UK families do not have this route at all.

**Claude (Anthropic): 18.** The consumer version is adults only. [Anthropic's help center](https://support.claude.com/en/articles/15171100-age-assurance-on-claude) says Claude "is only available to people over 18 years," and that it will "disable accounts based on indicators of minor activity."

**Character.AI: no open-ended chat for under-18s.** The company's [own help center](https://support.character.ai/hc/en-us/articles/42645561782555-Important-Changes-for-Teens-on-Character-ai) told teens there would be "no more open-ended chat for under-18 users" from November 24, and that it was adding "new age assurance functionality" using its own model plus a third party. Other creative features stayed. We looked at what that leaves a teenager with in our piece on [whether Character.AI is safe for kids](https://parents.theaieducator.io/posts/is-character-ai-safe-for-kids).

So there is no single AI age limit. There are four different rules, and a child moving between apps in one evening crosses all four.

## Does the age limit actually stop anything?

Mostly it decides which version of the product your child gets, rather than whether they get in at all. That is the practical answer, and it is changing.

Until recently, every one of these limits was self-declared: the child picks a birth year and the app believes it. That is still how most sign-ups work, which is why I keep saying to parents that **the age limit is a promise, not a lock**.

What is new in 2026 is that two of these companies have stopped relying only on the promise. OpenAI has [said it is building a system](https://openai.com/index/building-towards-age-prediction/) "to understand whether someone is over or under 18," and that "if we are not confident about someone's age or have incomplete information, we'll take the safer route and default to the under-18 experience." Anthropic now runs age checks through a third party, Yoti, when its systems flag an account, and says that "Anthropic never sees your ID or image; we receive only a pass/fail result."

Regulators are pushing the same way, and [California's new chatbot laws](https://parents.theaieducator.io/posts/california-ai-chatbot-law-for-kids) are the sharpest example of it so far. In a policy statement on February 25, 2026, the [US Federal Trade Commission](https://www.ftc.gov/news-events/news/press-releases/2026/02/ftc-issues-coppa-policy-statement-incentivize-use-age-verification-technologies-protect-children) told companies it would hold off on COPPA enforcement for those that adopt age verification, provided they "take reasonable steps to determine that any product, service, method, or third party utilized for age verification purposes is likely to provide reasonably accurate results."

My reading of that, and it is my reading rather than anyone's finding: the age question is moving from something your child answers to something the company guesses. That is probably safer. It also means a wrong guess now has consequences, and the most common wrong guess is a fifteen year old being treated as an adult because the account was set up years ago with a made-up birthday.

> An age limit does not decide whether your child gets to use ChatGPT. It decides which version of ChatGPT your child gets. And in most homes, your child is the one who chose.

## Why is the number almost always 13?

Because of a privacy law, not a judgment about maturity. In the United States, the [Children's Online Privacy Protection Act](https://consumer.ftc.gov/articles/protecting-your-childs-privacy-online) requires services aimed at under-13s to get a parent's consent "before they collect, use, or disclose a child's personal information." Thirteen is the age above which that obligation falls away. It is a data rule that the industry adopted as a de facto maturity rule, and the two are not the same thing.

This matters at home for one reason. When a company says 13, it is not telling you that 13 year olds are ready. It is telling you where its legal paperwork changes.

Worth knowing: most parents do not have this number. [Ofcom's Children and Parents report](https://www.ofcom.org.uk/media-use-and-attitudes/media-habits-children/childrens), published on May 21, 2026, based on 5,008 interviews with UK parents of children from six months to 17 and 3,426 interviews with children aged 8 to 17, found that "around a third (32%) correctly say the minimum age for most social media platforms is 13." The same report found 56 percent of UK children aged 8 to 17 have used AI, rising from 35 percent of 8 to 9 year olds to 66 percent of 16 to 17 year olds. So a majority of children are using this, and a minority of parents know the rule.

## What I See in Schools

This is where the age question stops being about your child's honesty and starts being about a setting in somebody else's console, and it is the part almost no guide for parents mentions.

I sit on the school side of this. I have run digital strategy in a college, I work with schools, trusts and districts on their AI strategy, and I have advised the Department for Education, KHDA in Dubai and the Ministry of Education in Kazakhstan. The pattern I see most often is this: a school buys or switches on an AI tool, and the age question is answered once, by an administrator, in a settings panel, and then never looked at again.

Here is the concrete version. On Google Workspace for Education, access to AI features is governed by an age label that the school sets. [Google's own admin guidance](https://knowledge.workspace.google.com/admin/getting-started/editions/control-access-to-google-services-by-age) tells the administrator, not the parent, to do it: "As an administrator, you should identify users in your organization who are 18 years of age and older." For a primary or secondary school the default is the cautious one, because "all users not designated as 18 and over get an age-restricted experience for some Google services." But a default is only a default. Someone can change it for a year group, for a class, or for a group of students who were designated once and have since moved on.

The same shape applies on the ChatGPT side. OpenAI's education guidance is explicit that for children under 13 in an educational setting, "the actual interaction with ChatGPT must be conducted by an adult." I have never once seen that sentence quoted in a school's AI policy.

Three things follow, and they are what I would actually want a parent to know.

First, your child may be using an AI tool at school that they are below the consumer age limit for, entirely lawfully, because the school signed a different agreement on your behalf. That is not a scandal. It is how school technology has always worked. It is just usually invisible to parents.

Second, the written AI policy and the admin console rarely match. The policy is written by a senior leader, often carefully. The console is configured by a network manager, often at speed, sometimes years earlier. When I ask a school which age category its students sit in, the honest answer is frequently that nobody in the room knows, and finding out takes a week.

Third, and this is the one that changes what a safeguarding lead can do: on a school account the school can usually see activity, and on your child's personal account at home it cannot see anything at all. Parents often assume the school's oversight follows the child. It follows the account.

## The Stated Age Check: the two minutes I would spend tonight

Every filter, every teen safeguard, every content restriction on these products hangs off one number, and in most families nobody has ever looked at it.

**The Stated Age Check is the two-minute thing to do tonight: open the AI app your child uses, find the account settings, and look at the date of birth it holds. Not the one you assume. The one that is actually in there. Everything the company does to protect your child, every filter, every control, every teen safeguard, hangs off that one number, and your child chose it.**

If the date is wrong, fix it together rather than quietly. A birthday changed behind a teenager's back gets discovered in about a day, and what they learn is to make the next account somewhere you cannot see.

If your child is on a school-issued account as well, that is a second, separate number, and you do not control it. Ask about it at parents' evening.

## How this differs by age

**Under 8.** No independent chatbot account. If they use AI it is on your device, with you there, as a thing you do together. Google's own line that Gemini "should not be treated as a friend" is written for exactly this age, where a child genuinely cannot tell the difference.

**8 to 12.** This is the band where the stated age usually gets invented, because most of what their friends are using says 13. If you want them to have access, the supervised route (a Family Link account with Gemini switched on, in countries where that exists) is the honest version of it, because it puts your name on the decision instead of a false birthday. Our [guide for 9 to 12 year olds](https://parents.theaieducator.io/posts/ai-for-9-to-12-year-olds) goes further into which account to use.

**13 to 15.** They meet the platform rule properly for the first time, and they also get the teen experience, which is a genuinely different product with more restrictions. The mistake here is assuming the teen version is supervised. It is filtered. Those are different things, and [ChatGPT's parental controls](https://parents.theaieducator.io/posts/chatgpt-parental-controls) still do not let you read their conversations.

**16 to 18.** The age limit stops being the useful conversation. What matters now is whether they can tell when a confident answer is wrong, and whether they would tell you if something they read there frightened them.

## What to say to your child

Short, unbothered, and not at bedtime after a row.

> You could say: "What birthday is on your ChatGPT account? Not a trick question, I just want to know which version you're getting."

> You could say: "Some of these apps say 13, some say 18, and school has its own thing. I'd rather sort out which one you're on than pretend I know."

> You could say: "If an app ever asks you to prove how old you are, come and get me before you send anything. That's my job, not yours."

The third one matters more than it looks. Age verification increasingly means a face scan or an ID, and a child alone with that prompt will either lie or hand over something they should not.

## What to ask the school

Three questions, and a good answer to the first one has a name and a system in it, not a policy quotation.

1. Which AI tools are switched on for my child's year group, and which age category are their accounts set to?
2. Who set that, and when was it last checked?
3. If my child types something worrying into one of them, who sees it, and how quickly?

If your child's school has not spoken to parents about AI yet, the honest place to start is a look at where it actually stands rather than a new policy document. That is the [school AI readiness work I do with schools](https://theaieducator.io/school-ai-readiness), and it is worth forwarding to a head of year or a principal.

## What none of this can guarantee

No age setting stops a child opening a browser and using a chatbot logged out. No filter catches everything, and Google says so itself. No parental control on any of these products lets you read your child's conversations. And nothing on this page overrides your school's own rules, a platform's stated age limit, or the advice of a doctor or a safeguarding lead if you are worried about your child rather than about a setting.

The number is worth knowing. It is just not the thing keeping your child safe. That is still the conversation, and the fact that asking is ordinary in your house.

If you want this kind of thing explained without the panic, I write a [Sunday newsletter](https://theaieducator.io/#newsletter) for teachers and parents. Parents are very welcome.

*Dan Fitzpatrick is a former secondary teacher, assistant headteacher and Director of Digital Strategy who now helps schools and families make sense of AI. [More about Dan](https://theaieducator.io/about).*

## Frequently Asked Questions

### What happens if my child put a false birthday on their AI account?

They will be getting the adult version of the product, with fewer content restrictions and no teen safeguards. Fix it with them rather than behind their back: change the date of birth in account settings together, and explain that the number decides which filters apply. Some services, including ChatGPT and Claude, now also try to estimate age from how an account is used, so a false birthday can later trigger a verification request.

### Does my child's school need my permission before letting them use an AI tool?

Often not, because in the United States COPPA allows a school to give consent on a parent's behalf for educational use, and in the UK schools rely on their own lawful basis. That is why a child below a tool's consumer age limit can use it lawfully at school. You can still ask which tools are switched on, which age category the account is set to, and whether there is an alternative. Our guide on asking to opt a child out of AI at school covers how to word that.

### Can I read my child's AI conversations if I set up parental controls?

No. OpenAI states plainly that parental controls do not let a parent or guardian read or monitor a teenager's conversations. Controls can restrict features, set quiet hours and send certain safety alerts, but the transcript is not shown to you on any of the major consumer products. If you need to know what your child is talking about, the conversation is the only route.

### Is there an AI chatbot that is genuinely built for younger children?

There are products marketed that way, and they help at the margins, but none removes the need for supervision, because a filtered chatbot only protects a child for as long as it is the app they happen to be in. Common Sense Media recommends no AI chatbots at all for children aged 5 and under. We looked at the whole category in our piece on whether a kid-safe AI chatbot exists.

### Are the age rules different in the UK?

Some are. The biggest practical difference as of September 2026 is that Gemini Apps are not available for supervised Google Accounts in the United Kingdom, the European Economic Area or Switzerland, so the under-13 supervised route that US families have does not exist here. ChatGPT, Claude and Character.AI apply the same stated ages. Ofcom reported in May 2026 that only around a third of UK parents correctly identify 13 as the minimum age for most platforms, so it is worth checking rather than assuming, and your child's head of year can tell you what the school has enabled.

### Should I just ban AI until my child is old enough?

A ban is a reasonable choice for younger children and a fragile one for teenagers, because a logged-out browser reaches most of these tools in seconds and a banned tool is one your child will not tell you about. For under-8s, supervised use together is the honest version. From about 11 upward, the thing that actually protects a child is that asking about it is ordinary in your house.