On July 1, Ohio became the first state where every public school district is legally required to have an AI policy. Oklahoma’s new law adds parental notification and an opt-out. Georgia, Maryland, and Idaho have their own mandates rolling in, and 31 states have published at least guidance. Here’s what none of that paperwork does: tell your students what’s allowed on your assignments.
District policies are written for school boards. The sentence that actually governs September — “can I use ChatGPT on this essay?” — gets answered by you, in your syllabus, or it gets answered by each student privately, in whatever way benefits them most. And there’s a second document almost nobody prepares: the note that reaches parents before the first AI dispute, so the first time they hear about AI in your classroom isn’t an accusation email.
Both documents take about ten minutes with AI’s help — which is either ironic or fitting, depending on your mood. Here’s the whole job: pick one of three rules, generate your syllabus statement, and send the parent email that prevents the fight.
Step 1: Pick your rule (there are only three)
Every workable classroom AI rule is a version of one of these. University teaching centers converged on the same three tiers — what Suffolk’s Center for Teaching and Learning publishes as model statements, Stanford’s CTL frames as “state whether, where, and how” — and they map cleanly onto K-12:
Two things make any of the three work. Per-assignment beats per-course: “Tier 1 for in-class essays, Tier 2 for research projects” is clearer than one blanket rule, and it matches how the university statements are actually written. And the why matters: research on academic integrity consistently finds students follow rules more when the reasoning is explained and revisited out loud, not just printed — a rule that says “because this assignment builds a skill AI would do for you” outperforms “because I said so.”
If you’re deciding under time pressure: Tier 2 as the default, Tier 1 explicitly named for in-class and skills assessments. That combination covers a typical grade 6–12 classroom and leaves you room to loosen later, which is much easier than tightening.
Step 2: Generate your syllabus statement (copy-paste)
Open ChatGPT (or Claude or Gemini — this works identically) and paste this, filling the brackets:
Write a classroom AI-use statement for my syllabus. Grade level:
[7th grade English]. My rule: [Tier 2 — AI tools are allowed for
brainstorming and checking grammar, but not for writing sentences
or paragraphs that get turned in. Any AI use must be listed at the
top of the assignment. In-class writing is always AI-free.]
Requirements:
- Under 150 words, at a reading level my students understand
- State the rule, then ONE sentence of "why" in plain language
- Include exactly how to disclose AI use (one line at the top of
the assignment: what tool, what for)
- State what happens the first time the rule is broken: a
conversation and a redo, not an automatic zero
- Tone: firm, calm, not scary. No legal language.
You’ll get something close to done. Edit two things by hand before it goes in the syllabus: make the consequence line match what you’ll actually do (never publish a threat you won’t keep), and read it once as your most rules-lawyering student would, looking for the loophole. Found one? Tell the AI to close it: “a student could argue [X] — rewrite the rule to cover that.”
The same prompt with [Tier 1] or [Tier 3] brackets produces the other versions — takes a minute per class if you run different rules for different courses.
Why the disclosure line is the load-bearing part: with a disclosure habit, AI use becomes visible and gradeable instead of secret. The students who’d never cheat get a legitimate path to tools they’re already using, and the conversation shifts from catching to coaching. That shift matters more than which tier you picked — especially since AI detectors can’t referee this for you. Detectors flag human writing as AI constantly (Stanford research measured a 61% false-positive rate on non-native English speakers’ essays), which is why universities from Berkeley to Washington State have stopped using them as evidence. We covered that full story — and what to do when you suspect AI anyway — in our detector false-positives guide.
Step 3: The parent email (the document nobody writes)
Here’s the fight this email prevents: a student turns in suspicious work, you raise it, and the parents hear about your AI expectations for the first time inside an accusation. Of course they push back — from where they sit, the rule appeared retroactively. Oklahoma’s new law now requires districts to notify parents about classroom AI use; whether or not your state does, the September version of this email costs you nothing and buys you the entire fall.
Paste this:
Write a short email to parents of my [7th grade English] students
about how AI is handled in my classroom this year. My rule:
[Tier 2 — allowed for brainstorming and grammar with disclosure;
never for writing that gets turned in; in-class work is AI-free.]
Requirements:
- Under 200 words, warm and matter-of-fact
- Lead with the positive: we teach responsible AI use, not fear
- State the rule in two sentences a non-teacher instantly gets
- One sentence on HOW I handle suspected misuse: I talk with the
student first and look at their drafts and process — I never
rely on an AI detector score as proof
- Invite questions now, before the school year gets busy
- Do not mention specific AI products by name
That fourth bullet is the sentence that ends arguments before they start. When a dispute does come, you’re no longer defending a surprise rule with contested detector evidence — you’re following a process every parent read in September: conversation first, drafts and process as evidence, no detector-score convictions. It also happens to be exactly the standard the research supports, and the one your district’s lawyers would pick anyway.
What this means for you
If you teach middle or high school English or social studies: you’re the front line — do all three steps this week while it’s quiet. Your subjects generate the most AI temptation and the most disputes.
If you teach elementary: the parent email matters more than the syllabus statement — your students’ AI use happens at home, on homework, often with a parent’s login. Adapt the email to “here’s how to help without the AI doing the work.”
If you’re a college instructor: same three tiers, but put the statement in the syllabus and on each major assignment — the university teaching-center versions (Stanford’s CTL has examples) assume course-level autonomy you actually have.
If your district already has a policy: yours can only be more specific, not more permissive. Read the district version once (Ohio’s model policy is public if you want the reference standard), then write the classroom translation of it. “The district allows AI with teacher discretion” still needs your discretion written down.
If you’re a homeschool parent or co-op lead: you’re both the teacher and the parent here — write the rule anyway. The clarity is for the student, and the three-tier model works at a kitchen table exactly as it does in a classroom.
What this can’t fix
- A rule doesn’t detect anything. It sets expectations and gives you standing when something’s off — the what happens when I suspect process still runs on drafts, conversation, and judgment, not scores.
- Ten minutes of policy doesn’t redesign your assignments. If an essay can be fully outsourced to ChatGPT, a syllabus line won’t stop that — we cover the redesign move in the companion piece on AI-resistant assignments.
- It won’t match every colleague’s rule. Students will have Tier 1 in your class and Tier 3 next door; that’s fine, and it’s another reason the rule has to be written down rather than assumed.
- It doesn’t decide your school’s stance on AI teaching tools — that’s the district policy’s job. This is only the piece of the mandate wave that lands on your desk.
The bottom line
The states are mandating policies; the classroom rule is still yours to write, and unwritten rules get litigated in October. Ten minutes: pick a tier, generate and hand-edit the syllabus statement, send the parent email before anyone needs it. If you want this exact workflow taught step by step — the three-tier rule, the parent email, the detector-free evidence standard, and the 90-second in-class check — our new course AI for Teachers: Your First Week Back starts free, and its two free lessons build the very documents in this post. For the broader teaching side — lesson planning, feedback, differentiation — Teaching with AI is the companion.
Sources
- AI Model Policy for Ohio Districts and Schools — Ohio Department of Education and Workforce
- Ohio’s July 1, 2026 School AI Policy Deadline — KJK
- 4 more states require districts to adopt AI policies — K-12 Dive
- Oklahoma SB 1734 — Oklahoma Senate
- Model syllabus statements for generative AI — Suffolk County Community College CTL
- GPT detectors are biased against non-native English writers — Stanford HAI
- WSU discontinues Turnitin AI detection — WSU Provost
- AI use in schools growing, but district policies haven’t caught up — GovTech
