AI by age: what's actually appropriate from 5 to 17

There is no single right answer, and that is the actual problem

Almost every article about children and AI is written as though all children are the same child. They are not. The advice that fits a sixteen year old doing NCEA is close to useless for a six year old asking why the sky is blue.

So here it is broken down the way New Zealand schools actually group children.

Primary, Years 1 to 6, roughly ages 5 to 10

The headline: no major AI platform supports this age group. ChatGPT requires users to be 13 or older. There are no parental controls available for a primary aged child, because from the platform's perspective that account should not exist.

Which means you are the safety layer. There is not another one.

What works at this age:

• Together, on your account, never on their own
• Short and purposeful. "Let's ask it about volcanoes" beats open ended browsing
• Good uses: curiosity questions, making up stories, explaining something they asked you and you could not answer
• The habit to build: "let's check that"

The main risk here is not rude content. It is the confident wrong answer. AI states nonsense with total authority, and an eight year old has no reason to doubt an adult sounding voice.

Intermediate, Years 7 to 8, roughly ages 11 to 12

The awkward middle. Most are still under 13, so still unsupported by the platforms, but homework has become real and the social pressure has started.

What changes:

• Homework support genuinely begins to be useful here
• Introduce the idea of saying where help came from
• Supervision shifts from sitting beside them to checking in with them
• Start asking them to explain their work back to you

The main risk is habit forming ahead of judgement. A child who learns at 11 that AI produces the answer will not unlearn it at 15.

College, Years 9 to 13, roughly ages 13 to 17

Now the platforms support them, and the assessment rules bind.

• Parental controls exist for 13 to 18 accounts. Quiet hours, voice off, memory off, Study Mode on by default
• You cannot read their conversations. That is deliberate design, not a setting you have missed
• For NCEA external assessment, generative AI is not permitted
• For internal assessment, your school's authenticity policy sets the rule

The main risk is outsourcing thinking during exactly the years thinking is being built. Not dramatic. Just quietly costly.

A way of thinking about it that already exists

We do not agonise over whether a film is "safe". We ask whether it is G, PG, M or R13, and we accept that the answer changes with age.

AI deserves the same framing. The question is never "is this safe". It is "is this appropriate for this child, right now, for this purpose". That is a question you are already qualified to answer.

Three things that do not change at any age

1. Check it. Always. Even at 17.
2. Be able to explain how you used it. If they cannot, they did not learn anything.
3. Talk about it. The families who handle this well are not the ones with the strictest rules. They are the ones where it gets discussed at dinner.

‍