How We Think About Introducing Kids to AI

This Is Not a Hypothetical

Whatever any of us think about the right age for a child to meet a language model, the question has largely been settled by circumstance. Common Sense Media's 2026 study found 86% of children report having used generative AI tools, with nearly a quarter using them daily.^[1] Roughly three in four teens have used AI companions specifically.^[2]

"Should kids use AI" is no longer the live question. "What are they using, and was it built for them" is.

Most of It Was Not Built for Them

Independent testing has been unflattering. Common Sense Media's risk assessments found that major general-purpose chatbots consistently failed to recognize and respond appropriately to mental health conditions affecting young people, concluding the risk was unacceptable for that use.^[3] Age gates on most of these products are trivially bypassed, and content filtering varies widely between them.

Regulators have taken notice. In September 2025 the Federal Trade Commission opened a formal inquiry into AI chatbots acting as companions, ordering seven companies to explain how they evaluate safety for minors, how they monetize engagement, and how they disclose risk to parents.^[4] Several states have since imposed specific obligations on companion chatbots, including disclosure that the user is talking to a machine.

The Companion Problem

The category that concerns us most is not homework help. It is companionship. Survey data indicates about one in three teens has turned to an AI companion to discuss something serious rather than talking to a person.^[2] The same research finds that children who use AI daily or weekly are more likely to report feeling lonely.^[2]

That last finding is correlational, and we want to be careful not to overstate it — it does not establish that AI use causes loneliness, and the arrow could plausibly run the other way. But it is exactly the kind of signal that should make anyone building a chatting character for children slow down rather than speed up.

What AI Literacy Actually Means

The constructive version of this conversation already exists. UNESCO's AI competency framework for students defines AI literacy across four dimensions — a human-centred mindset, ethics of AI, AI techniques and applications, and AI system design — at three levels: Understand, Apply, Create.^[5] The framing worth borrowing is its goal: children as responsible users and eventual co-creators, not passive consumers.

For young children, that translates into some unglamorous fundamentals well before anything resembling prompt engineering:

• Knowing when they are talking to a machine, and that it is a machine.
• Understanding that it can be confidently wrong.
• Understanding that it is not a friend, and does not know them.
• Knowing which questions belong to a person instead.

For Parents and Teachers, Right Now

The most useful thing an adult can do is unremarkable: sit with a child the first several times they use an AI tool, ask it something you already know the answer to, and let them watch it get something wrong. A child who has personally caught a model being confidently incorrect has learned more about AI than any explanation will teach them.