AI in education: why schools have a double job to do

When school leaders hear “AI and the AI Act”, they tend to think of tech companies, not their own staffroom. That is a misunderstanding. Article 4 of the European AI Act addresses all organisations that provide or use AI systems, and makes no exception for education. A school that uses AI tools, and by now virtually every school does, somewhere between adaptive learning software, plagiarism detection and the teacher generating quiz questions with a chatbot, must therefore ensure a sufficient level of AI literacy among its staff.

Worth being clear about from the start: article 4 does not prescribe a specific course, certificate or diploma. It asks for demonstrable, appropriate knowledge among the people working with AI. How you achieve that is up to you, in a way that fits your school’s context.

Why schools are no different, and yet they are

Legally, a school is simply an organisation using AI, just like a municipality or a plumbing firm. But in practice, a school has a double role that other organisations do not have:

Role 1: your own team

Teachers, support staff and school leadership use AI in their work. Think of lesson preparation with a chatbot, adaptive practice software deciding which exercises a pupil gets, or tools that help with marking. That use requires the same basic knowledge as anywhere else: knowing what AI can and cannot do, recognising when an answer is unreliable, and understanding what you do and do not enter into an AI tool. That last point is extra sensitive in schools: pupil data is personal data of minors, and the GDPR rightly sets high standards for it. A teacher who pastes a class list with care notes into a free chatbot has created a privacy problem, not a time saving.

Role 2: modelling it for pupils

Here a school differs fundamentally from any other employer: pupils are watching. How the team handles AI is education, whether you intend it that way or not. A teacher who says openly “I prepared this with AI, and this is how I checked it” teaches pupils more about responsible AI use than many a lesson about it. It works the other way too: a school that only bans AI for pupils while staff quietly use it all the time loses credibility and misses an opportunity.

Because pupils already use AI at home anyway, for homework, for chatting, for making images. The question is not whether they will encounter it, but whether anyone teaches them to approach it critically. School and parents stand side by side here; for the home front we wrote a separate guide on children and AI chatbots.

Extra point of attention: AI that decides about pupils

The AI Act is not only about literacy. Certain uses of AI in education count as high-risk under the Act, for example AI systems used to determine access to education or to evaluate learning outcomes. Those come with heavier requirements, mainly for the providers of such systems, but schools that use them also need to know what they are bringing in. Which is exactly why basic knowledge across the team matters: you can only ask a vendor critical questions once you know which questions exist.

Practical starting points for school leadership

You do not need to wait for a national framework or a perfect policy document. A school leader can do this within the current school year:

1. Map actual use. Ask the team anonymously which AI tools they already use, for lesson preparation, marking, communication. The results almost always surprise, and without this picture you are steering on assumptions.
2. Cover the basics for the whole team. Not just the ICT coordinator and the enthusiastic early adopters, but everyone who works or will work with AI: what it can do, where it goes wrong, what you never do with pupil data. One shared baseline also makes the staffroom conversation easier.
3. Agree rules for pupil data. A short, clear line: which tools are approved, what may and may not be entered into them, and who to contact when in doubt. Involve your privacy officer or DPO.
4. Choose a line on pupil use. What may pupils do with AI in tests and assignments, and above all: what do you teach them about critical use? A rule without an explanation backfires.
5. Record what you do. Who has been trained, when, and what is the agreement on refreshers? It does not have to be complicated; how to set this up practically is covered in Recording AI literacy.
6. Schedule repetition. AI changes by the quarter. One study day in 2026 is a fine start, but not an end point. Put AI literacy on the agenda as a recurring item, for instance linked to existing training days.

Start small, but start

The temptation is to wait: for the sector council, for the ministry, for a neighbouring school that has already figured it out. Understandable, but pupils and staff are using AI now. A modest, honest start, basic knowledge for everyone, clear agreements on pupil data, and an open conversation about what AI means in the classroom, is worth more than an ambitious plan that will be finished next year.

Looking for a concrete way to cover the basics for your team? The AI literacy course was built for staff without a technical background and ends with an exam and a verifiable certificate. With a team licence, school leadership can see exactly who has completed the course. Want to see for yourself first? Try the free module. And for the conversation with parents, there is the parents’ course AI and your child.