How to set clear AI expectations for your own classroom — and protect student data — when your school hasn't handed you a policy yet.

Students are already using AI. Many schools still have no clear classroom rules. That leaves teachers in a difficult position: ignore it, ban it, or build practical expectations of their own.

The case for acting is real — as of 2026, about one in five U.S. districts still has no AI guidelines, and barely a third of teachers say their school has formal rules for student AI use. But "acting" has limits worth naming up front. You can set classroom expectations immediately: how students may use AI on your assignments, how they disclose it, and what counts as misuse. Tool approval and student-data decisions are different — those usually should follow your school or district's process, not your own judgement alone. This guide covers both halves: the rules you can set today, and the privacy questions you should route through your institution.

A note on scope: the legal specifics below are for the United States and Canada, where most of our readers teach. The principles travel anywhere, but if you teach elsewhere, check your own jurisdiction's equivalents.

A good classroom AI policy isn't a ban or a free-for-all. The frameworks that work — from the U.S. Department of Education's school toolkit to widely-used scales built by assessment researchers — share the same handful of components.

The components every use policy needs:

Sample — read aloud in week one:

This four-level shape mirrors what real schools publish. One U.S. high-school handbook, for example, puts the responsibility squarely on the teacher to "provide students with notice of whether AI use is permitted on a particular assignment," and defines specific tiers like "AI-assisted editing." Borrow the structure; adjust the levels to your subject. And align with your district where you can: if there's any official guidance, your class policy should sit inside it, not contradict it.

The four-level shape above is a simplified version of a published framework worth knowing by name: the AI Assessment Scale (AIAS), developed by Mike Perkins, Leon Furze, Jasper Roe, and Jason MacVaugh. Its premise is simple — a binary "use it / don't use it" rule is too blunt, so it gives you a scale of permitted AI use that you set per assignment based on what you're trying to measure. The original scale has five levels:

(The authors released a revised version in 2025, refined for K-12 and reframing the top level as "AI Exploration." The core idea — pick a level per task by your learning goal — is unchanged.)

This isn't only theory. In one single-institution pilot, AIAS adoption coincided with fewer AI-related misconduct cases and improved attainment and pass-rate measures (the authors report a 5.9% rise in attainment and a 33.3% increase in module pass rates). Treat it as encouraging rather than conclusive — but the logic is sound: when expectations are explicit, far fewer students cross a line they didn't know was there.

The scale only clicks when you see it applied. Here's the same assignment set at different AIAS levels — once in language arts, once in math.

A persuasive-essay assignment (language arts):

A problem-solving assignment (math):

Notice that the higher levels don't mean "less learning" — they move what you're assessing from production toward judgement. Label the level on the assignment itself, and students never have to guess.

Here's the part most class AI policies skip entirely: the moment you put an AI tool in front of students, you're making a data-privacy decision, and you're doing it inside a legal framework whether you know it or not.

The one rule to remember:

This is not legal advice. Treat the rest of this section as a practical checklist of questions to raise with your school, district, or privacy officer — not as an interpretation you should make on your own. The laws below set the stakes for why that approval process exists.

The FTC has actually enforced this. In its order against the ed-tech company Illuminate Education — after a breach exposed the data of millions of students — the company was required to implement a data-security program and limit the data it collects and retains. Student data is regulated, and the regulators are watching.

Your classroom AI rules are a promise to your students: here's how you may use this tool, and here's how I'm protecting your information while you do. The first keeps learning honest. The second is mostly about routing data decisions through the people whose job it is to make them. You can set the rules today; the privacy half is a set of questions to raise, not calls to make alone.

Start with one paragraph in your syllabus: when AI is allowed, how students disclose it, what happens when they misuse it, and which tools are approved. Then add the privacy line: no student data goes into unapproved tools. That paragraph — said out loud in week one — gives your students something most schools still haven't.

📎 Free download: "Classroom AI Policy Starter Kit" — a ready-to-adapt policy + privacy checklist for your syllabus.

Disclosure: Hossein works in AI and builds AI-related products. AI by Age takes no AI-vendor sponsorships. Full disclosure on our About page.