Where Do Your Child’s AI Notes Actually Go?

A worksheet has a child’s name in the corner. A photo of their science notebook shows a school logo. An AI app offers to turn both into neat study notes in seconds.

That sounds useful. But before uploading the image, there is another question: Where will the child’s information go? The quality of an AI summary and the privacy of its input are two different decisions.

A study note can carry more personal information than we notice

The obvious details are names, addresses and phone numbers. The less obvious ones are school names, class groups, handwritten notes, faces in the background, and a timetable visible on the desk. Even a prompt about a friendship problem might reveal something deeply personal.

UNICEF’s guidance for parents on AI recommends reviewing what tools collect and encouraging children to pause before sharing personal details. That is more practical than telling a child that everything online is either safe or unsafe.

A local folder, a cloud account and an AI model are not the same thing

While building an AI notes tool called ScribeShot, I learned to separate three questions: Where are files saved? Where are they processed? And what happens to them afterward? I built that tool, so this is a design lesson from my own work, not a recommendation that children should use it.

A file can be saved on your laptop while a cloud AI service processes a copy. A local model can process information without sending the prompt to a model provider, yet those files may still be included in device backups or accessible to other users of the computer. Neither the word local nor private settles the question by itself.

The app’s current settings, provider policy, account type and school rules matter. Some services may retain uploads or conversations for different periods and purposes. Don’t assume that deleting a visible chat automatically deletes every stored copy.

Use a four-question check before uploading a child’s work

• Do we need the real document? Could a parent type a generic version of the maths question instead of sending a photographed worksheet?
• What can we remove? Crop out names, faces, school details and other identifying information; check for clues in the background.
• Where does the information go? Read the tool’s current privacy, retention and model-training settings, and check whether it sends data to other providers.
• Who controls it later? Look for a way to review, export and delete uploads. Check whether the service is appropriate for your child’s age and permitted by the school.

For example, if the assignment is “explain fractions using pizza slices”, a parent can ask AI that generic question without uploading a named school worksheet. The useful explanation may be almost identical, but the information shared is much smaller.

There is one complication: sometimes the image really matters

A child might need help interpreting a diagram, handwriting or a particular error in their own work. In those cases, a generic question could lose the essential context. Rather than banning the tool, decide whether that context is worth sharing, whether you can minimise it, and whether a trusted alternative would work.

UNICEF’s 2025 guidance on AI and children treats children’s data protection and privacy as design responsibilities, not a burden families can solve entirely by being careful. If an app makes these basic choices impossible to understand, that is a reason to use something else.

The first privacy lesson is a question, not a prohibition

For children aged roughly 6–12, I would keep uploads parent-led and teach one habit: before sharing a picture, note or question, ask “Does the tool need to know this about me?”

If the answer is no, remove it. If the answer is unclear, don’t upload it yet. We can still enjoy what AI does well while teaching children that their information is something worth protecting.

Related reading: What Parents and Children Should Know About AI Memory.