School District AI Policy Template

1.1 This policy governs the use of artificial intelligence by students, staff and contractors of [District], on district devices and accounts and on personal ones where the work is district work.

1.2 It is written to make AI usable rather than forbidden. Students will encounter these tools for the rest of their lives, and a district that only prohibits teaches nothing except how to hide. Where this policy says no, it says exactly what and exactly why.

2.1 Student personally identifiable information is not entered into any AI tool that has not been approved by [District] under a written agreement. This includes names, student identifiers, grades, attendance, discipline records, health and special education information, and anything else that would be an education record.

2.2 Under FERPA, a vendor may only receive education records where a recognised exception applies and the district retains direct control over the data and its further use. A free consumer tool with terms nobody negotiated does not meet that.

2.3 For any service used by students under thirteen, COPPA obligations attach and verifiable parental consent or a lawful school-consent route must be in place before use, not after.

2.4 Before approving any tool, [District] establishes in writing: whether student input trains the vendor's models, how long input is retained, which subprocessors receive it, whether a human at the vendor can read it, and how data is deleted when the contract ends.

3.1 Only tools on the [District] approved list may be used with students or with student data. The list names the tool, the grade bands it is approved for, and what it may be used for.

3.2 Staff may request additions. Requests are answered within [ten] school days, because an approval process slower than a lesson plan guarantees teachers use something else.

3.3 AI features that arrive switched on inside software the district already licenses are treated as new tools and reviewed before use, not grandfathered in because the vendor is familiar.

4.1 Each assignment states which of three categories it falls into, and the teacher sets the category in advance: AI NOT PERMITTED, AI PERMITTED WITH DISCLOSURE, or AI ENCOURAGED.

4.2 Where AI is permitted with disclosure, the student states what tool they used and for what. A brief note at the end is enough. Honest disclosure is never itself penalised.

4.3 Silence from a teacher defaults to AI NOT PERMITTED for graded work. Ambiguity is the district's failure rather than the student's, so a teacher who did not state a category cannot treat unstated use as a violation.

4.4 Students remain responsible for everything they submit, including its accuracy. Submitting AI output containing invented sources or false claims is an integrity matter whether or not AI was permitted.

5.1 No student will be disciplined, given a reduced grade, or recorded as having violated this policy on the basis of an AI detection tool's output alone.

5.2 Detectors return a probability, not a finding of fact, and their error rates are higher for students who are not native English speakers and for students who write in an unusual register. Treating that output as proof punishes exactly the students least able to contest it.

5.3 Where a teacher has a concern, the process is a conversation. Ask the student to talk through their draft, their sources and their choices. A student who did the work can almost always discuss it, and that conversation is both fairer and more reliable than any score.

5.4 A concern is documented with what the teacher observed, never with a screenshot of a detector percentage as the substance of the allegation.

6.1 Staff may use approved AI for planning, drafting, differentiating materials, and administrative work, with student personal information excluded under section 2.

6.2 AI does not determine a grade. It may support feedback, and the professional judgement and the final mark belong to the teacher, who must be able to explain them.

6.3 AI is not used to make or substantially determine decisions about identification for special education, discipline, placement, or promotion and retention. A human makes those decisions and records the reasons.

6.4 Illinois districts: from 1 January 2027, state law restricts AI in public-school teacher evaluations and requires disclosure and human judgement in evaluation decisions. Confirm your own state's position before AI touches any staff evaluation.

7.1 No AI system is offered or presented to students as counselling, therapy, or mental or behavioural health care, and no AI substitutes for a school counsellor, psychologist or social worker.

7.2 Nevada prohibits AI performing those functions and bars AI therapist representations, with civil penalties, and a growing number of states restrict AI in mental health services. Confirm your own state before any wellbeing tool is considered.

7.3 Any AI tool available to students that could receive a disclosure of distress must surface a route to a human and to crisis support. If a tool cannot demonstrate that behaviour, it is not approved for student use.

7.4 Staff who become aware of a student in crisis follow the district's existing safeguarding process. No AI tool changes who is responsible.

8.1 Where AI functions as an accommodation, including text to speech, speech to text, translation, summarising and organisational support, it is permitted and its use is not a disclosure requirement under section 4.

8.2 Nothing in this policy is applied so as to remove a tool a student depends on to access the curriculum. Where an assignment prohibits AI, the teacher provides an equivalent route for students whose accommodation includes these tools.

9.1 [District] publishes the approved tool list and this policy where families can find them, and notifies families before a new category of AI use begins with their children.

9.2 Families may ask what tools their child uses, what data those tools receive, and how to raise a concern. The answer is provided in plain language and in the family's language where practicable.

10.1 This policy is reviewed each semester and whenever the approved list, a vendor's terms, or state law changes materially. Owner: [name and role]. Last reviewed: [date].

10.2 This is a starting draft rather than legal advice, and adopting a policy is not the same as complying with anything. Have it reviewed by your district's counsel against your state's student privacy law and your existing agreements before adoption.