The short answer
- Vague policies fail. “Use AI responsibly” means nothing; students need to know exactly what’s allowed on this assignment.
- The clearest model is three tiers: AI-free, AI-assisted (with disclosure), and AI-encouraged — labelled per task.
- Always require disclosure. The habit of citing AI help is the most protective thing you can teach.
- Explain the “why.” Students follow policies they understand; tie the rules to learning, not just rule-following.
- Make it a conversation, not a decree — and grab our ready-to-use template to start.
Most classroom AI policies fail for the same reason: they’re vague. “Use AI ethically and responsibly” tells a stressed sixteen-year-old nothing about whether they can use ChatGPT on tonight’s essay. A policy that works is specific, explained, and — ideally — built with students rather than dropped on them. Here’s how to write one.
Why you’re probably writing this yourself
Because almost nobody has been given a policy to work from. Gallup’s survey of 2,232 US teachers found that fewer than one in five have received any formal guidance on using AI across their teaching tasks. Broken down, it’s worse in exactly the areas where judgement calls hurt most:
| Task | Teachers receiving no guidance |
|---|---|
| One-on-one instruction / tutoring | 69% |
| Grading and student feedback | 58% |
Meanwhile 60% of teachers are already using AI tools (30% weekly), and 54% of teens are using AI chatbots for schoolwork. So the policy vacuum isn’t theoretical — both sides of the desk are already using these tools, mostly without written rules.
There’s also an equity dimension worth knowing before you write: 67% of low-poverty districts have trained teachers on AI use, against 39% of high-poverty districts. If your school is in the second group, the document you’re about to write may be the only guidance your students ever get.
Why do most AI policies fail?
Because they answer a philosophical question when students have a practical one. Students don’t need your stance on the future of AI; they need to know, for this assignment, what’s allowed. Ambiguity is where the trouble lives: a student who genuinely doesn’t know whether AI help is permitted will either avoid a useful tool or misuse it, and either way feels unfairly caught when the grade comes back. Specificity, per task, removes almost all of that.
The three-tier model
Label every assignment as one of three levels. It’s the clearest framework available, and it fits on a slide:
| Tier | What it means | Use it for |
|---|---|---|
| 🚫 AI-free | No AI use at all | Assessing individual skill; in-class work; foundational practice |
| ⚙️ AI-assisted | AI allowed for specific, disclosed purposes | Most homework — e.g. explaining concepts, checking work, feedback on drafts |
| ✅ AI-encouraged | Students expected to use AI and reflect on it | Tasks about using AI well; advanced or authentic projects |
Put the tier on the assignment itself. When students never have to guess, most policy disputes simply don’t happen.
What every policy should contain
Four things, in plain language:
- What’s allowed and what isn’t — specific to your class, ideally per assignment via the tiers.
- How to disclose — a standing expectation that students note any AI help. Keep it light: a sentence is usually enough. (Point them to how to cite ChatGPT for when you want it formal.)
- Consequences — what happens if the policy is broken, kept proportionate and focused on learning.
- The reasoning — why these rules exist, tied to the fact that school is building skills the students themselves will need.
Why explaining the “why” matters
Students follow rules they understand and route around rules they don’t. A policy that says “no AI on this essay because I need to see that you can build an argument, which is a skill the AI can’t have for you” earns compliance that “no AI, per school policy” never will. You’re not just regulating behavior; you’re teaching why the behavior matters — which is the actual job.
Involving students helps even more. A ten-minute conversation about what feels fair — where AI genuinely helps learning versus where it replaces it — produces a policy they own, and ownership beats enforcement every time.
Put it to work
Pair your policy with assignments designed to be AI-resistant, so the rules are backed by tasks where following them is the path of least resistance. Then start from our classroom AI policy template — adapt the wording to your subject and voice (a policy that reads like it came off the internet gets treated like it came off the internet), and you’ll have something workable in ten minutes.
Related: rubrics in the AI era · the parent AI-policy letter.
Sources
- Gallup / RAND American Teacher Panel — US K-12 teacher survey (n≈2,232) on AI adoption, weekly use, and the share receiving formal guidance, including the breakdown by task and the district-poverty gap. news.gallup.com
- Pew Research Center — US teens (13–17) and AI chatbot use for schoolwork. pewresearch.org
Figures as of August 2026.
Frequently asked questions
What should a classroom AI policy include?
At minimum: which uses of AI are allowed and which aren’t, how students should disclose any AI help, what happens if the policy is broken, and — crucially — why the rules exist. The most workable policies label each assignment as AI-free, AI-assisted with disclosure, or AI-encouraged, so students never have to guess.
What is the three-tier AI policy model?
It’s a simple framework where every assignment is marked as one of three levels: AI-free (no AI use, for assessing individual skill), AI-assisted (AI allowed for specific purposes with disclosure), or AI-encouraged (students are expected to use AI and reflect on it). Labelling each task removes ambiguity, which is where most policy problems come from.
Should students have to disclose when they use AI?
Yes — disclosure should be a standing expectation. It teaches academic honesty, protects students from false accusations by documenting their process, and gives you insight into how AI is being used in your class. A simple line (‘I used ChatGPT to check my outline’) is enough; it doesn’t need to be a formal citation every time unless you require one.
How do I get students to actually follow the policy?
Explain the reasoning, keep it specific per assignment, and involve students in the conversation rather than handing down rules. Policies students understand and had a voice in are followed far more than ones that appear as fine print. Revisiting it after the first assignment — what worked, what was unclear — also helps a lot.