The AI Policy Is Adopted. Does Anyone in the Building Know What It Says?

Ohio’s deadline was July 1, 2026. Every public school district, charter school, and STEM school in the state had to adopt a policy on the use of artificial intelligence, either the state’s model policy or one built to align with it. Idaho, Maryland, Oklahoma, and Virginia followed with their own laws this year, each one requiring school boards to adopt a policy that matches new state guidance. More states requiring an AI policy for schools are moving through statehouses right now.

So the paperwork question is mostly answered. If you run a school district, you likely have an AI policy on file, adopted, dated, filed with the state, checked off. The harder question is the one nobody put a deadline on: does anyone who works in your buildings actually know what it says?

Which states require an AI policy for schools right now

Ohio and Tennessee moved first, requiring districts to adopt their own AI policies. This year, Idaho, Maryland, Oklahoma, and Virginia went further, passing laws that direct state education departments to publish AI guidance and require every school board to adopt a matching local policy. The details differ by state, and the differences tell you where the real work is.

Maryland’s law requires districts to name a coordinator responsible for the “productive and ethical use” of AI, a real person with a real job description, not a policy sitting alone in a binder. Oklahoma’s law goes the other direction, restricting AI from being the primary basis for grading or discipline decisions, and giving parents the right to opt a student out of AI use in school without academic penalty. Ohio published a model policy districts could adopt wholesale or customize. None of these laws tell a district how to make the policy real once it is signed.

That gap is where most districts are standing right now, holding a compliant document and an open question about what happens next. More states will land in that same spot before the school year is out. Watching the leading edge now, instead of scrambling once your own statehouse acts, is the cheaper path.

A model policy is not a used policy

A policy that lives in a school board packet is invisible to the people it is supposed to guide. The classroom teacher deciding whether to let a student use AI for a first draft has not read the seven-page board policy. The paraprofessional fielding a parent’s question about an AI tutoring app does not know the opt-out language exists. The business office staffer evaluating a new AI vendor has no idea the policy names a specific coordinator to loop in first.

None of that is a failure of the law. The laws did what laws do: they set a floor. They did not, and could not, make a document into a habit. That part was always going to be the district’s job, and it is the part almost nobody built a plan for, because the deadline pressure went toward getting something adopted, not toward making it stick.

This is the same pattern behind almost every AI rollout that stalls. A tool gets approved. A policy gets signed. A training gets scheduled once. And then the actual behavior in classrooms and offices keeps running on whatever staff already believed about AI before any of it happened, because nobody made the new rule visible at the moment it mattered.

What makes a policy visible instead of filed

The districts getting this right are not doing anything exotic. They are doing three specific things that a compliance memo will not do on its own.

First, they put the policy where staff already look. Not a standalone PDF on the district website, but a line in the new-hire packet, a slide in the back-to-school staff meeting, a laminated one-pager at the front office next to the fire evacuation route. Visibility is a placement decision as much as a communication one.

Second, they name a person, the way Maryland’s law now requires. Whether or not your state mandates a coordinator, someone in your district needs to be the answer to “who do I ask.” A policy without an owner turns every edge case into a guess.

Third, they revisit the policy on a cadence, not just at adoption. AI tools change month to month. A policy written in early 2026 will need language for tools nobody had heard of by the time this school year ends. Districts that treat the policy as a living document, reviewed twice a year, catch that drift. Districts that treat it as a box checked in July do not.

What compliance actually requires beyond the document

Here is the uncomfortable part for a business office already stretched thin: the law asks for a document, but the actual risk you are managing, the FERPA question, the equity question, the “did a teacher just let AI grade an IEP-supported student’s essay” question, lives downstream of the document, in a hundred small decisions made by people who never read it closely.

That is not a reason to write a longer policy. It is a reason to spend less time on the policy itself and more time making sure the people covered by it actually understand what it means for their specific role. A ten-minute conversation with the front office about the opt-out process does more for real compliance than another revision of the policy’s legal language.

The states writing these laws are, in effect, telling every district the same thing school finance offices have known for years: a rule that exists on paper and a rule that shapes behavior are two different achievements, and only one of them protects a district when something goes wrong. Auditors do not ask whether a policy exists. They ask whether anyone followed it, and whether the district can show its work.

If your district has a policy on file and no plan for how staff actually learn it, that is the gap worth closing before the next state adds itself to this list. The Spotlight Machine’s AI training sessions are built for exactly this handoff, taking a district’s adopted policy and turning it into something staff can actually apply on a Monday morning, not just cite in a board packet.