Writing an AI Policy for Your School
A practical framework for writing a fair, workable AI policy — for a classroom, a department or a whole school — with the questions to answer and language you can adapt.
Start with purpose, not prohibition
The instinct when writing an AI policy is to reach for a ban, but a policy built only on "don't" ages badly and tells students what to avoid without telling them why. A better starting point is the question the policy actually needs to answer: when does using AI support a student's learning, and when does it replace the learning the assignment was meant to produce? That distinction — support versus replacement — is the spine of every workable AI policy, and it gives students a principle they can apply rather than a list of rules they will look for ways around.
What a Department-Wide AI Ban Does and Does Not Settle
A department-wide ban sets one clear expectation for students across courses, but the reports on Trinity College do not describe how violations would be detected. According to the Trinity Tripod, the college's student newspaper, the philosophy department voted unanimously to adopt a blanket restriction. The policy states: "Deliberate use of AI in philosophy courses, in any capacity and for any purpose, is forbidden."
The Tripod reports that this prohibition covers summarizing readings, brainstorming, research for presentations or oral exams, and writing or editing support. In an article for Daily Nous, Justin Weinberg reports that Prof. Shane Ewegen said the department wanted one blanket rule rather than a set of individual course policies that contradict each other. A Daily Nous update notes that the policy applies to all current philosophy faculty, while new hires are invited to help revise it.
The Tripod also reports statements from Ewegen regarding the broader campus environment. According to the report, Ewegen said the college administration publicly endorsed AI, purchased an AI software suite for all students, and did not plan to treat student AI use as a serious infraction. Ewegen said he led a faculty petition for a college-wide ban.
Neither the Tripod nor Daily Nous describes how faculty would detect or enforce violations. Because the reports are silent on enforcement, we do not know whether Trinity College's ban is enforced or how it performs in practice. Trinity College's campus-wide Policy No. 11.4, effective April 1, 2025, offers separate guidelines for instructors rather than an enforcement mechanism for the philosophy department.
Policy 11.4 instructs faculty to "Clearly communicate expectations to your students in all of your courses regarding permitted and restricted uses of generative AI tools and include such statements in class syllabi." The same document advises instructors to "Exercise caution when using generative AI plagiarism detection tools, as their accuracy is not guaranteed; such tools can return false positives and introduce bias against non-native English speakers, as well as students with disabilities or diverse learning needs." The reports do not state that the philosophy department relies on detection software.
In our reading, a ban provides a single standard for students, but any school adopting one must resolve the missing procedural rules. We suggest pairing any restrictive policy with four practical safeguards:
- A clear scope list defining whether the ban applies to summarizing, brainstorming, research, drafting, spell-checkers, search engines, or accessibility tools like screen readers.
- A transparent disclosure norm that tells students when and how to report permitted technical assistance.
- A reliance on process evidence, such as version histories and project notes, rather than automated scoring tools. You can read more about checking for AI use without a detector to see how process evidence works.
- Supervised checkpoints, including in-class writing sessions and brief oral follow-ups. When questions arise, consult our guide on whether AI detectors are accurate and follow our steps for when a student is falsely accused of using AI by starting with an open conversation instead of an automated penalty.
This case is limited to one college department at a single institution, so it does not predict how a ban functions across an entire K-12 school or district. Our view would change if published records document Trinity College's enforcement methods or reveal student outcomes under the rule. For an analysis of district-level restrictions, see our discussion of the New York City and Los Angeles public school tool bans later on this page.
Based on our reading of the Tripod, Daily Nous, and Policy 11.4, a prohibition establishes a clear boundary for coursework, but it leaves daily enforcement unresolved. Before you vote on or publish a departmental ban, write these four evidentiary standards directly into your draft policy.
The questions a good policy answers
Whether you are writing for one classroom or a whole district, a usable AI policy answers a short list of concrete questions:
- Where is AI use allowed, limited, or prohibited? Be specific by task: brainstorming and studying may be encouraged, while generating submitted work is not. Many policies use a simple tiered model — green (use freely), yellow (use with disclosure or in specific ways), red (do not use).
- How must students disclose AI use? A simple, consistent disclosure expectation — a note describing what AI was used for — normalizes honesty and removes the ambiguity that drives a lot of misuse.
- How does AI relate to academic integrity? State plainly that submitting AI-generated work as one's own is a form of academic dishonesty, and connect it to your existing integrity policy rather than creating a separate, harsher regime.
- What guidance do teachers get? A policy that tells students the rules but leaves teachers to design assignments unaided will not hold. Include guidance on AI-resistant assignment design.
- What about privacy and appropriate tools? Address student data — students should not enter personal or identifying information into AI tools — and note any tools the school has vetted or restricted.
Prefer principles to product lists
AI tools change month to month; a policy that names specific products will be out of date before the school year ends. Describe categories of use instead — "generative writing tools," "AI tutoring and study aids," "AI detection" — so the policy still makes sense as the specific tools shift. Principles about disclosure, integrity and appropriate use age well; a list of named apps does not.
Write it with the people it governs
A policy handed down from the front office without teacher input tends to sit unread. The policies that actually change behavior are written with the people who live under them — teachers who know what is realistic in a classroom, and, where possible, students and families who will follow and support them. A small working group that drafts, tests the language against real assignments, and revises produces something teachers will actually use, because it reflects the classroom rather than an abstraction of it.
Enforcement without an arms race
Here is the hard truth every policy has to reckon with: AI detection tools are not reliable enough to be the enforcement mechanism. They produce false positives, they disproportionately flag English learners, and several institutions have disabled them for exactly those reasons — a subject we cover in depth in our guide on whether AI detectors are accurate. A policy that leans on detection to catch cheaters is building on sand. The durable alternative is to enforce through design and disclosure: assignments that make misuse pointless (in-class writing, process artifacts, oral defense), a clear disclosure norm, and integrity conversations rather than gotcha accusations. Enforcement becomes less about catching students and more about making honest work the natural path.
Lessons for Policy Writers from the New York City and Los Angeles AI Bans
New York City's rules, announced Sept. 2, 2026, ban AI-powered learning tools for elementary and middle school students, according to EdWeek. They also prohibit chatbots that offer mental health or emotional support for all students, and they prohibit school-issued computing devices in pre-K through 2nd grade.
Our guide to which tools New York City flagged lists all 38 products with the status Chalkbeat reports for each.
High schools get five pilot programs that allow limited use of vetted AI tools. The pilots cap use at five classes per school, a set number of minutes, or a restricted number of assignments. Generative AI features in roughly 40 other tools will be turned off on school devices, and the district will drop any tool whose AI cannot be disconnected.
Teacher guidance did not change. Teachers may not use AI to determine grades, but they may use it with caution for lesson planning, caregiver emails, and analyzing student data.
The ban applies for the 2026-27 school year, and the district will assess its impact. The sources reviewed give no exact end date. EdWeek Market Brief reports that the city "has a year to figure it out" and decide which guidelines replace the ban. At a Sept. 2 meeting of its generative AI ad hoc committee, Los Angeles Unified announced a moratorium on AI use on district-owned devices for all students this school year, EdWeek reports.
For a school or department policy:
- Write the replacement plan in the same document as the end date. In your own policy, name who runs the review, what they measure, and the date they hand over successor rules. The AI adoption checklist for schools lists questions to settle before a tool is approved.
- Name the students who are exempt. New York City exempts students in special education and English learners who need AI or technology to access the curriculum. Write that exception into your policy, and decide who approves it.
- Pair a tool ban with a plan to teach safe use. Richard Buery, CEO of the Robin Hood Foundation, was quoted by EdWeek Market Brief as saying the city "painted too broad a brush" and called for an "AI driver's license" so students learn safe, responsible use. His advice was "Move slow, move with caution, but move." If you restrict tools, schedule the lessons that replace them using the guide to teaching AI literacy. If your rules also cover chatbots, read Chalkbox's guide to the Common Sense Media rating of ChatGPT for Teens before you draft that clause.
Keep it short, clear and revisable
The best AI policies are short enough that students and parents actually read them, written in plain language rather than legalese, and explicitly marked as a living document you will revisit as the technology and your experience evolve. A one-page policy that everyone understands beats a ten-page one no one finishes. Plan to revise it at least yearly.
Turn the policy into practice
A school-level policy sets the frame; individual teachers make it real in their courses. The most direct way to do that is a clear statement on each syllabus — our AI syllabus statement examples offer copy-ready language for allowing, limiting or prohibiting AI in a specific class. Pair the policy with guidance on using AI well, like our overview of how to use AI in the classroom, so the message students hear is not only "here is what you can't do" but "here is how AI can genuinely help you learn." A policy that only restricts will be resented; one that also guides will be followed. When a school filter blocks a tool, our guide to how to request an AI unblock covers the written request to send IT.
Revisit it as you learn
The single most useful attitude toward an AI policy is to treat it as a draft you will improve. The technology, the tools students use, and your own understanding will all change over a year, and a policy frozen in place quickly stops matching reality. Build in a yearly review, invite teachers to flag what is and is not working, and adjust. A living policy that evolves with real classroom experience earns far more respect — and compliance — than a static document handed down once and never revisited. The aim is not a perfect policy on day one, but a fair, clear one that gets better as your school learns.
For the companion-chatbot question students and parents raise most, see the safety risks of Character AI for students.
Join the Chalkbox list for free printable packs and new tools — no spam, unsubscribe anytime.
Frequently asked questions
What should a school AI policy include?
Where AI use is allowed, limited or prohibited; how students must disclose AI use; how AI relates to academic integrity; guidance for teachers on assignment design; and a note on privacy and appropriate tools. Clarity matters more than length.
Should schools ban AI entirely?
A school or department can ban AI, and Trinity College's philosophy department did so per the Trinity Tripod, but a ban alone does not tell you how violations will be found. Published reports on the Trinity decision do not describe how the policy is detected or enforced. Most workable policies separate uses that support learning from uses that replace it, establishing clear expectations for each course.
How would a department enforce a ban on AI use?
Reports on Trinity College's philosophy ban in the Trinity Tripod and Daily Nous do not say how violations would be detected or enforced, so a department needs its own plan. Trinity College's Policy No. 11.4 cautions that detector accuracy is not guaranteed and can return false positives. In our reading, a ban should pair with disclosure, process evidence, supervised or oral work, and checking for AI use without a detector so concerns start with a conversation before an accusation.
Who should write the policy?
Ideally a group — teachers, administrators, and where possible students and families — so it reflects real classroom practice and has buy-in. A policy handed down without teacher input rarely sticks.
How is a policy enforced fairly given unreliable detectors?
By focusing on disclosure, process and integrity rather than on catching students with detection software. Assignment design and clear expectations do more than enforcement tools that produce false positives.
Does the policy need to name specific tools?
Better to describe categories of use than to name tools, which change constantly. Principles age well; a list of named products does not.
Do schools have to have an AI policy?
Not everywhere, but the list of states requiring one is growing. Ohio's House Bill 96 made it the first state to require every district to adopt an AI policy, with a July 1, 2026 deadline. Several other states passed similar mandates the same year. Confirm the current rule with your own state education agency or district office, and write one even where it is not required, since it is worth doing regardless.
Should the policy also address teachers' own use of AI, not just students'?
Yes. A policy that only restricts students while staying silent on teacher use of AI for grading, feedback or lesson drafting looks one-sided and invites the same disclosure and accuracy questions in reverse. Cover both in the same document, even briefly, so the standard is consistent.
What is an example of an AI policy?
Our AI syllabus statement examples guide gives copy-ready language you can adapt for a single class. Use it alongside the framework above when you need a school- or department-wide policy.
Is there a free AI policy template for schools?
Yes, TeachAI offers a free AI Guidance for Schools Toolkit. Developed by groups including Code.org, Khan Academy, and the International Society for Technology in Education (ISTE), the toolkit provides sample policies, planning frameworks, and communication materials. It is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0, which allows schools to adapt the text for non-commercial use. A downloaded template is only a starting foundation, so staff must tailor the text to local classroom needs and review it with the people it governs before adoption.
This guide is general information for educators, not legal advice. AI tools and school policies change quickly — verify specifics against your own school’s rules and the tools’ current documentation before acting.