Most pupils have already met artificial intelligence long before a school teaches them anything about it. They use it to draft, to summarise, to answer homework questions and to settle arguments. The question facing schools is no longer whether to introduce AI, but how to teach pupils to use it well, and how to prove they have actually learnt to.
This guide sets out what AI literacy means at KS2 and KS3, why it has moved up the agenda so quickly, and the practical decisions a school faces when it starts teaching it.
Why AI literacy matters now
Two things have collided. The first is adoption. Surveys of secondary pupils suggest a clear majority have already used generative AI, many of them regularly, and often with no guidance on how to use it responsibly. The tools arrived in children's hands faster than any curriculum could respond.
The second is policy. Following the national curriculum review, AI and digital literacy are set to be written into a new computing programme of study. That turns AI literacy from an optional extra into something schools will be expected to teach and to show they are teaching. Schools that start now are simply ahead of a change that is coming regardless.
What AI literacy actually means
It helps to be precise, because "AI literacy" is often used loosely. It does not mean teaching children to use a particular chatbot, and it is broader than online safety.
AI literacy is the ability to understand what these tools are, to question what they produce, and to use them thoughtfully and safely. A pupil who is AI literate knows that a model can be confidently wrong, can spot when an answer needs checking, understands why their data matters, and can tell the difference between using a tool to think and using it to avoid thinking. Those are transferable habits of mind, not features of one app that will look dated in a year.
AI literacy at KS2
At primary level the goal is foundational and age-appropriate. Younger pupils can grasp, in concrete terms, that a computer can make things that look real but are not, that it can be wrong, and that some information should never be typed into it. The emphasis is on curiosity and caution in equal measure: encouraging children to be interested in what AI can do while building an instinct to question it.
This is also the phase most often left out. Many of the well-known resources start at secondary, which leaves primary schools improvising. Building AI literacy at KS2 means children arrive at secondary already asking the right questions rather than starting from zero.
AI literacy at KS3
By KS3 the same ideas can be taught with more depth. Pupils can look at how models are trained and why bias creeps in, examine when AI use is appropriate for a piece of work and when it undermines the point of it, and start to articulate their own judgements. The honest framing many teachers land on is not "don't use AI" but "use it, and be able to say what you did and what you thought yourself".
The value of teaching this as a progression, rather than a one-off assembly, is that pupils build on what they learnt at KS2 instead of covering the basics again.
The hard part: evidencing learning, not just usage
Here is the challenge schools keep arriving at. Getting pupils to use AI is easy; they already do. Proving that a child has actually learnt to use it well is much harder. A completed activity does not tell you whether the understanding stuck, and "we did an AI day in the autumn term" is difficult to stand behind if a parent, an inspector or a governor asks what pupils can now do.
This is where AI literacy differs from a lot of edtech. The point is not the content alone, which is increasingly easy to find, but the evidence: a record of what each pupil has understood, a way to see progression across a year group, and something a school can point to as proof. A school that can show its pupils' AI literacy, rather than assert it, is in a far stronger position than one that cannot.
How schools are approaching it
Three practical decisions tend to come up first.
The first is policy. Schools want a clear position on where AI use is acceptable, where it is not, and how staff should respond, so that pupils are not left guessing and teachers are not each inventing their own rule. A simple, well-communicated policy does more good than a long one nobody reads.
The second is teaching. Schools decide whether AI literacy sits inside computing, is spread across subjects, or runs as a dedicated strand. There is no single right answer, but the schools that make progress tend to pick one owner rather than leaving it to fall between departments.
The third is safeguarding. AI literacy and online safety overlap, and the strongest approach treats them together: teaching pupils not just how to use AI, but what to be wary of, what to share and what never to.
Getting started this autumn
The simplest way to begin is to teach AI literacy as a short, structured programme with a clear progression from KS2 to KS3, and to build in a way of evidencing what pupils have learnt from the outset rather than bolting it on later.
That evidence layer is the part schools most often wish they had built in from the start, and it is exactly what we designed Curio to provide: self-marking modules pupils work through, a dashboard that shows a teacher what each class has understood, and certification a school can hold up as proof. Schools starting this autumn get their first six weeks free, which is a low-risk way to begin ahead of the curriculum changes. If that is useful, we would be glad to talk it through.