What should an AI syllabus statement say?

It should name where AI is allowed, where it is not, and what happens on a violation, in plain language a student can follow. The strongest statements distinguish by activity rather than by tool: AI is welcome for understanding, brainstorming, and feedback, and not for producing the graded work itself. A blanket ban is neither followed nor fair, and a blanket allowance abandons the point of the assignment.

Should teachers ban AI in the syllabus?

A total ban rarely holds. It is unenforceable across phones and home devices, and it removes a genuinely useful study tool from the honest majority. A clearer approach permits AI for learning support and prohibits it for generating submitted work, so students know the line and you can defend it consistently.

How do you enforce an AI policy on a syllabus?

On school-managed devices, enforce it where the work happens: allow AI for tutoring and explanation, block the generation of whole assignments, and keep a log of attempts. Off managed devices, enforce it through assignment design, in-class writing, and oral defense. The syllabus states the rule; the classroom and the device settings make it real.

An AI Syllabus Statement That Actually Holds Up

A syllabus statement that just says "no AI" does not hold up, because you cannot enforce it and you would not want to if you could. What holds up is a statement that draws the line by activity: AI is welcome for understanding and feedback, and it does not belong in the production of the work you are grading. That distinction is enforceable, it is fair, and students can actually follow it. Here is how to write one, with language you can lift.

Why the one-line ban fails

"Use of AI is prohibited in this course" feels clean and lasts about a week. It fails on three fronts at once.

It is not enforceable. Students have phones and home laptops you do not control. The ban does not stop AI use, it just moves it out of sight.
It is not fair. A blanket ban punishes the honest student who wanted to ask a clarifying question and does nothing to the one who quietly generated the whole essay at home.
It throws away a real tool. Used well, AI is a patient tutor that explains a concept five different ways at midnight. Banning the category to stop the misuse is like banning calculators to stop cheating on arithmetic.

The alternative is not "anything goes." It is a line drawn in the right place.

Draw the line by activity, not by tool

The durable distinction is between using AI to learn and using AI to submit. Say which activities are welcome and which are not.

Generally fine, because they build the student:

Asking AI to explain a concept, a reading, or a piece of feedback.
Brainstorming directions, then choosing and developing your own.
Getting feedback on a draft you wrote, the way you would from a tutor.
Checking your understanding by having it quiz you.

Not acceptable, because they replace the student:

Generating the essay, the code, the problem set, or the analysis that you submit as your own.
Producing text you then present as your original writing.
Using AI on assessments meant to measure what only you can do unaided.

Students can hold that line in their heads. "AI can help me understand, it cannot do the graded work for me" is a sentence a fourteen-year-old can follow.

A statement you can adapt

Here is a version to make your own. Keep it in plain language and put it near the top of the syllabus, not buried in a policy appendix.

On using AI in this course. AI tools can be a genuinely helpful way to learn. In this class you are welcome to use them to understand ideas, get unstuck, brainstorm, and get feedback on work you have written yourself. What you may not do is have AI produce the work you turn in. The essays, code, and answers you submit must be your own thinking, in your own words. If you are ever unsure whether a use is okay, ask me first, and you will never be penalized for asking. When AI helped you learn something, a quick note of how you used it is welcome, not held against you. Submitting AI-generated work as your own is treated as plagiarism under our integrity policy.

Adjust the consequences line to match your school's existing integrity language so it is consistent, not a separate regime.

Make the statement real

A syllabus statement sets expectations. It does not, by itself, enforce anything. Two things make it real.

Assignment design. Weight the work toward things AI cannot do for a student invisibly: in-class writing, drafts with visible history, and assignments defended out loud. When the thinking has to happen in front of you, the line enforces itself.

Device governance, where you have it. On school-managed devices, you can enforce the exact distinction your syllabus describes. AI stays available for tutoring and explanation, generating a whole assignment is blocked with a clear message, and attempts are logged with context. The student who wanted help gets it. The student who wanted a ghostwriter hits a wall, and you have a record instead of a suspicion. The syllabus says the rule; the device settings and your classroom make it true.

The point

A syllabus statement is a promise about how your class will treat a powerful, permanent new tool. Ban it and you make a promise you cannot keep. Draw the line by activity and you make one you can: help is welcome, doing the work for you is not, asking is always safe. That is a statement that holds up in week one and in week fifteen, and it is the foundation everything else in your integrity approach can stand on.