The Faculty Lounge
A school-wide artificial intelligence policy you can adapt, written for K-12 administrators. Fill in the highlighted parts, cut what does not apply, and take it to your leadership team.
Before you start
This is a template, not legal advice. Run the final version past whoever handles your student records and your legal counsel, because privacy obligations vary by state and by contract.
[District or School Name] uses this policy to set expectations for how artificial intelligence is used by students, staff, and administrators. The goal is to let people use these tools where they genuinely help, while protecting student data, academic honesty, and the professional judgment of educators.
This policy is written on the assumption that AI is already in the building. It is not a decision about whether to allow it.
This policy applies to [students, staff, contractors, volunteers, and third-party vendors] when using generative AI tools for any purpose connected to school work, whether on school devices or personal ones, on or off campus.
It covers text generators, image and video generators, coding assistants, AI features built into tools you already use, and any tool that produces content from a prompt. It does not cover long-standing automated features such as spellcheck, autocomplete, or grammar suggestions.
These commitments shape every rule below.
A note from The Faculty Lounge
The principle that does the most work here is the first one. Most of the real risk in school AI use is not students cheating. It is an adult letting a tool make a call about a child, then not being able to explain how the call was made. Write that line down and it settles a dozen future arguments.
Students may use approved AI tools when a teacher has said an assignment allows it. Typical permitted uses include brainstorming, generating practice problems, explaining a concept a different way, checking their own work before submitting, and studying.
Permission is set per assignment, not per class or per year. The default is that AI is not permitted unless the teacher says otherwise.
Staff may use approved AI tools for drafting materials, generating differentiated versions of existing resources, building rubrics and comment banks, drafting family communication, summarizing documents they have the right to read, and administrative writing.
Staff remain responsible for anything they send or post. Read every word before it goes out.
Administrators may use approved AI tools for drafting communication, translating family-facing material, summarizing meetings, and routine operational writing. Any output that will be sent to families or the public is reviewed by a person first.
Do not build this policy on AI detectors
Detection tools are not reliable enough to carry a discipline decision, and they fail unevenly. In peer-reviewed testing, seven widely used detectors flagged 61 percent of essays by non-native English writers as AI-generated while performing near-perfectly on native-speaker writing. Vanderbilt disabled its detector and published the arithmetic: at a claimed one percent false positive rate across the 75,000 papers it processed in a year, around 750 papers could have been wrongly flagged. Michigan State and Johns Hopkins made the same call, and OpenAI withdrew its own detector for low accuracy.
If your policy names a detector as evidence, you will eventually discipline a student who did nothing wrong, and it will most likely be an English learner. Use conversation, drafts, process, and the student's own account instead.
Undisclosed AI use on an assignment that did not permit it is handled under the existing academic integrity policy at [reference your policy number]. It is not a separate offense category.
Teachers state AI permissions on the assignment itself, in writing. Where AI is permitted, students add a short disclosure naming the tool, the stage of the work, and what they asked it to do.
First instances are treated as a teaching moment with a conversation and, where appropriate, a redo. Escalation follows the existing integrity policy.
A note from The Faculty Lounge
A policy students have never read is not a policy. It is important that an educator explains the AI policy to students in language those students can understand. See the individual policies by division level.
Only tools on the district's approved list may be used with anything connected to a student. The approved list is maintained by [role or office] and reviewed [how often].
Before a tool is approved, [role or office] confirms that the vendor's terms meet applicable requirements under FERPA, COPPA, and [your state student privacy law], and that student data will not be used to train the vendor's models.
Most consumer AI tools set a minimum age of 13 and expect parental consent through 17. For students below that age, AI use is teacher-operated: the adult runs the tool, and students do not hold accounts.
AI tools that support students with disabilities, including text-to-speech, reading support, and writing assistance, are permitted as accommodations and do not require separate disclosure when they are named in an IEP or 504 plan.
If an assignment requires AI use, the school provides access during the school day on school equipment. No assignment may require a subscription or a personal account.
All staff receive training before the start of [school year] covering the approved tool list, what may not be entered into a tool, how to set assignment-level permissions, and how these tools fail.
Students receive age-appropriate instruction on AI use during [when and in which course or advisory]. This is instruction, not a signature on a form.
A note from The Faculty Lounge
A nationally representative survey by RAND, published in September 2025, found that more than 80 percent of students said their teachers had never explicitly taught them how to use AI for schoolwork, even though a majority were already using it. RAND's December 2025 follow-up put student use at 62 percent, up from 48 percent seven months earlier. Make sure your community understands the policy.
Suspected misuse is reported to [role]. A suspected academic integrity violation begins with a conversation with the student before any determination is made, and no determination rests on detection software alone.
A suspected data exposure, meaning student information entered into an unapproved tool, is reported to [role] within [timeframe] and handled under the existing data incident procedure at [reference].
This policy is reviewed at least annually by [role or committee], and sooner if the tools or the legal landscape change materially. Last reviewed: [date].
One more thing
Adopt the shortest version of this that your school will actually follow. A two-page policy that principals can use beats a twelve-page one that no one bothers to read.
Free from The Faculty Lounge. If you want help adapting this for your school, our consulting team works with schools on exactly this.