The question that comes before all the others
"Can my child use AI?" has no single answer, because the assistants are not the same thing. One requires you to be 18 and offers no controls for minors at all. Another has the most developed parental control system on the market. And one of them simply cannot be turned off.
This guide is the map: which assistant allows your child, from what age, and what you can actually control in each. It is different from the site's other two AI guides — this is the overview; for setting up ChatGPT step by step, and for virtual companions, there are dedicated guides linked at the end.
Quick comparison
| Assistant | Minimum age | Parental controls | What you can do |
|---|---|---|---|
| ChatGPT | 13 | Yes, built-in and complete | Quiet hours, filters, study mode, risk alerts |
| Claude | 18 | None exist | Nothing: it is not built for minors |
| Gemini | 13 | Yes, through Family Link | Turn it on or off, plus a first-use email |
| Meta AI | 13 | Partial | See topics and get alerts. Cannot be turned off |
| Copilot | 13 (varies by country) | Indirect | Block the app or the site, and limit time |
ChatGPT: the most complete
It allows users from 13 and has gone the furthest. Since August 2026 it has a specific teen experience that applies on its own: the system estimates a user's age and, if it calculates they are under 18, moves them into a protected version regardless of what they entered. On top of that you can link your account and control hours, filters, memory, and study mode.
Claude: no minors allowed
There are no nuances or settings here. Anthropic's terms require users to be at least 18, or the legal age in their country if higher. There is no teen profile, no parent dashboard, and no way to link accounts. If your child uses it, they are outside the terms of service.
It is worth saying plainly, because there is a common assumption that every assistant matches ChatGPT. Not so: the only way to control Claude is from the phone, blocking the app and the site with Screen Time or Family Link.
Gemini: controlled through Family Link
Google allows access for under-13s on a supervised account, but you have to enable it — it ships off. The path is Family Link > Controls > Gemini > Gemini Apps. You get an email the first time your child activates it, and you can switch it off whenever you want.
Two important limits. First: you cannot read their conversations. The control is on/off, not supervision, and the keep-activity setting does not even exist for under-13s. Second, which matters in Europe: Gemini is not available for supervised accounts in the European Economic Area, Switzerland, or the UK. There, a young child simply cannot use it this way.
For younger kids, Google applies extra filters that stop Gemini from claiming to be a person, claiming human emotions, or role-playing a harmful character. Even so, Google itself warns that the filters are not perfect, and recommends explaining to your child that it is a tool, not someone to confide in.
Meta AI: the one you cannot switch off
It lives inside WhatsApp, Instagram, Messenger, and Facebook — and that is the problem: there is no switch to disable it. Meta said as much when announcing its teen controls: the assistant stays available, with age-appropriate protections. What you can do, if you have Instagram supervision turned on, is see the topics your teen asked about over the last seven days, and get an alert if they discuss suicide or self-harm with the assistant. What you will never see is the messages.
One thing worth knowing: in January 2026 Meta globally paused teen access to its AI characters while it rebuilds that feature. The pause covers both those who stated they are teens and those Meta suspects are. The general assistant still works; the characters do not.
Copilot: no switch of its own
Microsoft allows users from 13, with variations by country worth checking. But there is no Copilot-specific control inside Microsoft Family Safety: you have to block it manually as an app and as a website, separately, on every device.
What independent reviewers say
Common Sense Media, which evaluates digital products for kids, rated Gemini's under-13 and teen versions as high risk, and its conclusion applies to the whole category: they look like adult versions with safety filters bolted on, not products built for kids from the ground up. In 2026 it went further and rated Google's AI search answers an unacceptable risk, partly because they appear on their own on school computers with no way to turn them off.
Its age guidance is sensible and easy to remember: no one under 5 should use chatbots; ages 6 to 12 only with an adult present; 13 to 17 independently, but for schoolwork and creative projects. And one line that holds at any age: no one under 18 should use them for emotional support or companionship.