You can block a named AI app on your child's Android phone in about five minutes, in Google's Family Link app. The Chromebook is where it gets complicated. If your child signs in to that laptop with the account their school gave them, your settings do not apply at all: the school's administrator decides what AI they can reach, and you are not in that arrangement.
Blocking an AI app on Android tonight
Start here, because it is quick and it works. (If the phone in question is an iPhone, the route is different, and I have written that one up separately.) In the Family Link app on your own phone, tap your child, then Screen time, then Time limits, then App limits. Find the app in the list, tap it, and turn off the Allowed toggle. Tap Done.
Google's own Family Link guidance on managing a child's apps, read on September 25, 2026, sets out what happens next. The app "will be blocked or unblocked in about 5 minutes, or once the device is connected to the internet." If your child is using it at that moment they get "a 1 minute warning to finish up before the app is blocked." And usefully, the block travels: the same page says the app "should be blocked on all of your child's Android devices or Chromebooks," so one action covers the phone and any Chromebook signed in with that supervised account.
Two limits are stated on the same page, and they matter more than they look. "System apps can't have app limits placed on them." And "Some apps can't be blocked because they're required for parent supervision settings to work, or for your child's device to work." A system app is one that came with the phone rather than one anybody installed: the browser, the keyboard, the assistant that answers when you hold the power button. That is where a growing amount of AI now lives, which is why the app list on its own is not the whole job.
You could say: "I've switched that one off for now. You're not in trouble. I want to understand what you were using it for before I decide what the rule is."
Gemini sits behind its own switch
Google's assistant is not in the app list with everything else. It has a separate control, and a parent who blocks three AI apps and stops there will often find Gemini still running.
In Family Link, tap your child, then Controls, then Gemini, then Gemini Apps, and turn it off or on there. Google's guidance for parents on Gemini Apps, read on September 25, 2026, is unusually frank about what that switch does not do. "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." And: "Gemini is new and evolving, and can hallucinate and present inaccurate information as factual."
That second sentence is worth translating. To hallucinate, here, means to state something invented in the same confident voice it uses for something true. Ask it when your child's school was founded and it may give you a year, a founder and a short history, all of it plausible and some of it made up. It is not lying, because it does not know. It is guessing what a good answer would look like.
The same page confirms that a parent can switch Gemini Apps on for a child under 13, at a parent's discretion. So this is a decision, not a default you have to accept, and it is worth making it having read what Gemini is actually rated for.
Why your settings stop at the school's sign-in screen
Here is the part almost no guide tells parents, and it is the reason an evening of careful settings can come to nothing.
Parental controls follow the account, not the laptop. On a Chromebook, how much control you keep depends entirely on how the school's account was added to it. Google's Chromebook Help page on adding a school account for a Family Link user, read on September 25, 2026, draws the line in one sentence: "If you add a school account as a new user on your Chromebook's sign-in screen, Family Link parental controls will not apply to that user account."
Add the school account inside your child's supervised profile, as a second account, and your rules hold: "Screen time limits, downtimes, and other parental controls apply whenever a child is signed in." Add it as its own user on the sign-in screen, and you have created a second computer inside the same machine, with none of your settings on it.
On the school's own device, the question is settled before you open it. The same page states that "Google Workspace for Education administrators determine which Google services their users can access while signed into a Google Workspace for Education account." Google's page on managing a child's Chromebook, read the same day, repeats it and adds the consequence: a school account may carry "some features or services that your child was previously unable to access using the Family Link supervised account."
Signing in with the school's account, in other words, can hand back the very things you switched off.
So, before you change anything:
The Sign-In Check is the two-minute thing to do before you touch a single setting: open your child's laptop and look at the sign-in screen before anyone types a password. Count the accounts on it. As far as your controls are concerned, each one is a separate computer, and the evening you spend in Family Link only covers the one with your name behind it.
What I See in Schools
I have spent years on the other side of this, as a director of digital strategy in further education, and since then working on AI strategy with schools, trusts and districts. So I can tell you what that administrator's screen actually looks like, because it is the thing parents are never shown.
It is a list. A member of staff, usually one person with a dozen other jobs, decides for every pupil at once whether a Google service is on or off, and can set it differently for eleven-year-olds and sixteen-year-olds. When a new AI feature arrives, it frequently arrives switched on, and somebody has to notice and decide. The pattern I see most often is not a school making a bad call. It is a school making a real call, in a meeting, with reasons, and then telling nobody outside the building.
That is my reading of it rather than a finding from research, and I would say the same thing to the school. But it changes what a parent should do with an evening. The settings at home are worth twenty minutes. The question to the school is worth more, and almost nobody asks it.
Parental controls follow the account, not the device. Your child can be at the same kitchen table, on the same laptop, and be under a completely different set of rules depending on which account they signed in with.
What to ask the school
Keep it short and specific. A question about "AI" gets a policy. A question about a switch gets an answer.
- Which Google AI services are turned on for my child's year group right now, and who made that decision?
- If the school issues a device, is it still managed by the school when my child brings it home, and does that management cover AI tools as well as websites?
- If my child types something worrying into a school tool, what happens, and who sees it? (What a school can and cannot see is worth knowing before you ask.)
- Does the school's filtering and monitoring cover AI chat, or only websites?
That last one is the question I would push hardest on. Filtering built to block websites does not always treat a conversation inside an app the same way, and the honest answer from a school is sometimes "I will find out," which is a perfectly good answer.
If what comes back troubles you, there is a narrower request worth making, and what you can actually ask for is a shorter list than most parents expect.
How this changes by age
Under 8. Block the apps, switch Gemini off, and do not feel you owe an explanation beyond "that one's for older kids." At this age the bigger risk is the assistant that answers instantly on a smart speaker, not an app your child chose.
8 to 12. This is the age to do it together rather than overnight. Sit with them, show them the Family Link screen, and let them see you make the decision rather than discover it. If your child has a school Chromebook, this is also the age to look at the sign-in screen with them and ask which account they use for what.
13 to 15. Blocking alone starts to fail here, because a blocked app on a supervised phone is an invitation to use an unsupervised one. Keep the controls, and move most of your effort to knowing which tools they actually use. Ask what the last three things they asked it were. A general question gets a shrug; a specific one usually gets a phone held up.
16 to 18. Controls are mostly the wrong tool now, and a sixteen-year-old with a school account and a personal phone has more AI than you can switch off. What still works is the standing agreement: which tasks they will not hand over, and the fact that they can tell you when something a chatbot said unsettled them.
What blocking cannot guarantee
Three things, plainly.
It does not reach inside other companies' apps. Internet Matters, the UK online safety organization, states it in its guide to Family Link, last modified on July 15, 2026: "Google Family Link cannot restrict content within non-Google products." The AI inside Snapchat is a good example, and what My AI does and does not do is a separate decision from anything you set in Family Link.
It does not cover a signed-out browser. Nothing stops a child opening a chatbot's website without logging in to anything, on any machine.
And it does not reach school. That is the whole point of the section above, and it is not a loophole your child found. It is how the account was built.
None of this makes the block pointless. It makes it the first layer of three. The account decides what the app allows, the device decides what the screen allows, and the conversation decides everything else, which is why the last one is the only layer that covers the cases the other two miss.
What I would do this week
The numbers suggest the conversation is the layer most families are thinnest on. In the Common Sense Media census of AI use by tweens and teens, fieldwork run by SSRS from March 18 to 26, 2026 with 1,204 US children aged 9 to 17, 73% of children said their school or a teacher had told them what they should and should not use AI for in schoolwork, while 56% said a parent had talked to them about using AI safely. Those are children's reports of two different conversations, not a measure of how good either was. But the gap runs the direction most parents would not guess.
So: twenty minutes in Family Link tonight, including the Gemini switch. Then look at the sign-in screen on every laptop in the house, and find out which account your child is actually using for schoolwork. Then send one email to the school with the first question above in it.
If your child's school has not spoken to parents about which AI services are switched on, that is the work I do with schools, reviewing what is enabled and what the policy actually means in practice. It is worth sending to your head of year or principal. I also write a Sunday newsletter that keeps parents and teachers up to date without the panic.
Dan Fitzpatrick is a former secondary teacher and assistant headteacher who now helps schools and families make sense of AI.
