A parent’s guide to AI apps on kids’ tablets

If you hand a child a tablet today, there is a good chance they will meet AI on it, whether or not you installed anything that calls itself an "AI app". Homework helpers, drawing tools, games with chatty characters, keyboards that suggest replies and search features that summarise answers all increasingly have a language model behind them.

This is not a reason to panic, and a lot of it is genuinely useful. It is a reason to know where AI is, what it does with what your child types or says, and which settings put you back in control. This guide is written for parents who are not technical and do not want to become technical.

Where AI hides in kids’ apps

AI features tend to show up in a few familiar shapes:

• General chat assistants. The obvious ones. Many have minimum age requirements in their terms, commonly 13 and sometimes higher, which children routinely click past.
• Companion and character apps. Apps where the child chats with a persistent "friend" or character. These are designed to be engaging, sometimes very much so, and deserve the closest look.
• Homework and tutoring tools. Photo-a-question helpers and explainers. Useful when they teach; less useful when they simply hand over answers.
• Creative tools. Image generators, story makers and music tools.
• Features inside ordinary apps. A summarise button in a browser, an assistant inside a messaging app, smart replies on the keyboard.

The last category is the easiest to miss, because the app itself is one you already approved.

What happens to what they type

Most AI features do not run on the tablet. When your child types a question, speaks to a character or uploads a photo of their worksheet, that content is usually sent over the internet to a service that runs the model. What happens next depends on the provider: how long it is kept, whether it is used to improve the product, and who can review it. Those details are in the privacy policy, and they vary a lot.

A few things are worth teaching early, in plain words:

• AI chats are not private diaries. Assume someone could read them later.
• Do not share your full name, school, address, passwords or photos of yourself or friends.
• AI can be confidently wrong. It is a starting point, not an authority.
• If a chat ever makes you uncomfortable, stop and tell a grown-up. You will not be in trouble.

Companion apps deserve extra care

Of all the shapes AI takes on a child’s device, companion apps are the ones most worth slowing down for. They are built to feel like a relationship: the character remembers details, responds warmly and is always available. For an adult that can be pleasant; for a child or a lonely teenager it can become the place they go instead of to people. Some services have added age checks and safety features after public concern, but approaches vary widely between apps. If one of these is installed, look at it together, ask how it makes them feel, and decide as a family whether it stays.

Settings that help

You do not need special software to make a real difference. The built-in controls on both major platforms cover most of it.

On iPad and iPhone

• Screen Time → Content & Privacy Restrictions lets you require approval for app installs and limit apps by age rating.
• Ask to Buy, through Family Sharing, sends every download request to your device for approval, including free apps.
• App Limits can cap time in specific apps or whole categories.

On Android

• Google Family Link lets you approve app installs, set content ratings for Google Play and set daily limits.
• Check each app’s own settings for chat features, which are often separate from the app’s main function.

Inside the apps themselves

Many apps now have their own family or teen modes, the option to turn off AI features, or to stop conversations being used for training. They are worth five minutes per app, once.

Read the age rating, then read past it

App store age ratings are a helpful filter, but they describe the app’s content at review time, not necessarily what an AI feature might generate later. For anything with open-ended chat, read a few reviews from other parents, try it yourself for ten minutes, and check whether it has a way for you to see conversation history.

Talk about it like any other tool

The most effective control is still a conversation. Children who understand why the rules exist are more likely to follow them when you are not watching, and more likely to come to you when something feels off. A few questions that tend to open the topic without a lecture:

• "Which of your apps can talk back to you?"
• "Has the AI ever told you something that turned out to be wrong?"
• "What would you do if a chat character asked you something personal?"

Agree a simple family rule, such as AI apps only in shared rooms, or only for named tasks, and revisit it as they get older.

Seeing the whole house

Settings on one tablet do not tell you what the games console, the smart TV or the old phone in a drawer are doing. That is where a network view helps. GreyArm’s DNS shield runs on a computer in your home and, once your router points devices at it, shows which devices look up AI services and when. You can block specific AI services for specific devices, without installing anything on the tablet itself and without reading anyone’s messages.

It is not a replacement for conversations or for the built-in controls. It is a way to notice when something new shows up, so the conversation can happen early.

A five-minute checklist

• Turn on install approvals (Ask to Buy or Family Link).
• Set age ratings for the app store.
• Open each chat-capable app and check its family, history and training settings.
• Teach the four rules: not private, no personal details, can be wrong, tell a grown-up.
• Agree where and when AI apps are used.