Article 4 of the EU AI Act says it plainly. Providers and deployers of AI systems must ensure their staff have a "sufficient level of AI literacy."
That is not guidance. It is law.
The European Commission has gone further. In early 2026, the AI Office ran a series of regulatory webinars and published a repository of acceptable literacy practices. They made one thing very clear: you cannot tick this box with a single training module. The measures have to be tailored to the workforce, the use-case, and the population affected.
One-size-fits-all does not count. A compliance checkbox does not count. You need to show that the people using AI in your organisation actually understand what they are using.
Why this matters for schools
Schools in EU member states are not directly regulated under Article 4 in the same way as a bank deploying an AI credit scoring system. But the direction of travel is clear. If employers are legally required to upskill their staff on AI, schools are being asked the same question from the other direction: are you preparing students who will walk into those workplaces?
Every member state has to establish a national AI regulatory sandbox by August 2026. The compliance infrastructure is being built now. The curriculum expectations will follow.
Schools that treat AI literacy as optional today will find it is mandatory tomorrow. Not because Ofsted said so. Because the law did.
The US is moving too, just differently
The US does not have a single federal AI literacy law. What it has is chaos. Productive chaos, but chaos.
Idaho enacted a statewide K-12 framework that mandates local AI policies, literacy standards, and teacher training. It specifically bans AI from replacing human teachers. California passed legislation through its first chamber focused on student data privacy, making it illegal to use student data to train AI models. Illinois gave families the right to opt out of AI grading. Utah embedded AI ethics into mandatory Year 7 and 8 computer science standards. North Carolina allocated $100,000 in recurring annual funding for AI literacy in secondary schools.
Every state is doing something different. Some focus on ethics. Some focus on privacy. Some focus on curriculum. The Bipartisan Policy Center warned in March that without federal coordination, the country risks a patchwork that serves no one well.
But the signal is the same everywhere: AI literacy is moving from "nice to have" to "required."
What schools should take from this
Three things.
First, this is not slowing down. The regulatory machinery is moving fast. PISA 2029 will test AI literacy. The EU has made it law. US states are legislating term by term. The window for "we'll deal with it eventually" is closing.
Second, compliance is not the same as competence. A school that ticks a box saying "we covered AI in PSHE" has not built AI literacy any more than a company that puts everyone through a 45-minute e-learning module. The EU has been explicit about this. Measures must be contextual, specific, and demonstrable.
Third, the burden falls on every subject, not just Computing. The frameworks coming out of the OECD, the Council of Europe, and the EU all say the same thing: AI literacy is interdisciplinary. It belongs in History and Maths and English and Science and PE. The teacher who says "that's not my department" will find that it is.
AILitKit was built for exactly this moment. When a PE teacher needs to show how fitness tracking connects to algorithmic decision-making. When a Geography teacher needs to link data visualisation to AI bias. When a school leader needs evidence that AI literacy is being delivered across the curriculum, subject by subject, not in a single assembly.
The law is arriving. The frameworks are written. The question is whether your school is ready or still waiting for someone to tell them to start.
Matthew Wemyss is the founder of AILitKit and IN&ED, and author of AI in Education: An Educator's Handbook.