Six states now require every school district inside them to adopt a written AI policy. Four of those laws passed in 2026.
A school AI policy is a district-level rule set naming which AI tools staff and students may use, for which tasks, and with what disclosure. Read the new state laws next to each other and they converge on one line: AI can help produce and adapt teaching material, but it cannot decide a grade, a punishment, or a placement.
That leaves a wide permitted zone, and it is the zone Jellypod works in. A teacher uploads their own slides, readings, and notes, reviews the script the tool drafts, and publishes a short episode to a class feed. The teacher picked the sources and signed off on the script, and no student outcome was decided along the way.
Permission is not the problem. Most teachers have never been told what their district actually allows. A Gallup and Walton Family Foundation survey of 2,069 public K-12 teachers, published May 26, 2026, found that only 18% had received formal guidance of any kind from administrators on how AI tools should be used.
What is a school AI policy?
It is the document that answers three questions for your building: which tools are approved, which tasks they are approved for, and what has to be disclosed to students and families.
Most policies written in the last two years share a spine. They name an approved tool list. They require a human to make any decision that affects a student's record. They ask staff and students to attribute AI-generated text, images, and multimedia in submitted work. Some add a data clause covering what student information may be entered into a third-party system.
The variation is in enforcement. A policy adopted because the state told the district to adopt one tends to be specific and auditable. A policy adopted voluntarily tends to be a set of principles.
Which states require districts to have an AI policy?
Six, as of this school year. Idaho, Maryland, Oklahoma, and Virginia all passed laws in 2026 directing their state education departments to develop AI guidance and requiring school boards to adopt matching policies, according to K-12 Dive's July 9, 2026 roundup. Ohio and Tennessee got there first.
| State | What the law requires | Timing |
|---|---|---|
| Ohio | Every traditional district, community school, and STEM school adopts an AI policy | Deadline was July 1, 2026 |
| Tennessee | District AI policy aligned to state guidance | Start of the 2024-25 school year |
| Maryland | District policy plus a designated AI coordinator; state creates evaluation rubrics and professional development | 2026 law |
| Oklahoma | District policy; bars AI as the primary tool for grading or discipline; parental opt-out; annual disclosure of AI tools in use | 2026 law |
| Virginia | District policy aligned to state guidance, plus a funded pilot for instruction, tutoring, and teacher support | 2026 law |
| Idaho | District policy aligned to state guidance | 2026 law |
Far more states have published advice than have imposed a rule. As of October 2025, 34 states and Puerto Rico had released AI guidance for schools, per AI for Education's tracking cited in the same K-12 Dive report. Most of it is advisory. The six states above made writing a policy mandatory.
What do school AI policies actually prohibit?
The prohibitions cluster around decisions, not tools.
Oklahoma's law is the clearest version. It bars AI tools from being used as the primary means of grading, discipline, or other consequential decisions about a student. It lets parents opt their child out without academic penalty. It requires districts to disclose, once a year, which AI tools are in use and what student data they collect.
District policies written this summer echo the same structure. The recurring phrase is human in the loop, meaning the AI output is an input to a person's judgment rather than a replacement for it.
What almost none of these policies prohibit is using AI to prepare material. Drafting a study guide, differentiating a reading for three levels, or turning a lecture into something a student can listen to on the bus falls outside the banned category, because no student outcome is being determined.
Can teachers still use AI to make classroom materials?
Yes, and that is the use case the policies are written around rather than against.
The safe pattern has three parts. The inputs are material you already trust, usually your own. You review the output before a student sees it. And you disclose that AI was involved, in whatever form your district asks for.
Teachers were already working this way. RAND's 2025 panel report found 53% of English language arts, math, and science teachers using AI for school work, concentrated in planning and preparation rather than direct instruction or grading. The policies caught up to the practice.
Where it gets uncomfortable is the tool nobody approved. A policy that names an approved list means a tool absent from that list is a question for your coordinator, not a judgment call at 10pm. Maryland's AI coordinator requirement exists to give that question somewhere to go.
Why do most teachers still have no guidance?
Because writing a policy and teaching a staff are different projects, and the second one is slower.
The Gallup survey is blunt about the size of the gap. Fielded February 9 to March 2, 2026, it found 69% of teachers had received no guidance at all on using AI for one-on-one instruction or tutoring, and 58% had received none on grading and feedback. About half of those who did get something described it as informal.
So for most teachers the live question is not whether a tool is allowed. It is who can tell them, and in four of the six states the answer is still being assembled while classes are already running.
Where does audio fit inside a policy?
In the least contested part of it, which is teaching material built from sources the teacher already trusts.
Educators have been building this way on Jellypod without any policy telling them to. Excluding the default category assigned to brand new accounts, Education is the single largest category of shows created on the platform, and across roughly 700 education shows started in the last six months, about one in five is set to private or unlisted rather than published to a public directory. These are class feeds. They were never meant for an audience beyond the roster.
That shape matters for a policy conversation, because two of the recurring district concerns disappear. The content came from the teacher's own material rather than from an open-ended model prompt. The audience is the class rather than the open internet.
The workflow is short. Upload the readings or slides, read the drafted script and change what is wrong with it, generate the audio, and share the feed with your students. Our guides on repurposing lectures into podcasts and the best AI tools for teachers cover the mechanics. For the evidence behind whether any of this helps students, what the research says about AI in the classroom is the better starting point.
Frequently asked questions
Does my district have to have an AI policy?
If you teach in Idaho, Maryland, Ohio, Oklahoma, Tennessee, or Virginia, yes. State law requires your board to adopt one. Everywhere else it is currently optional, though 34 states and Puerto Rico had issued advisory guidance as of October 2025.
Can teachers use AI to grade student work?
Not as the deciding factor, under the stricter policies. Oklahoma's law specifically bars AI from being the primary tool for grading or discipline. Using AI to draft feedback that you then read, edit, and stand behind is a different act, and most policies treat it that way. Check your own district's wording before you rely on that distinction.
Do students have to disclose when they use AI?
Under most district policies, yes, and the disclosure usually covers text, images, and multimedia rather than writing alone. The specific format varies by district, so the answer lives in your policy document rather than in a general rule.
Can parents opt their child out of AI tools?
In Oklahoma, yes, and without academic penalty. Other states have not written the same provision, though individual districts have adopted opt-out language on their own.
Is using an AI tool to make a class podcast against school AI policy?
Almost never, when you supply the sources and review the output. These policies restrict AI that decides things about students. Making listening material from your own vetted lesson content is the kind of use the human-in-the-loop language is written to permit.
The short version
The 2026 school AI policies ban AI from deciding grades, discipline, and placement. They leave the preparation of teaching material open, on the condition that a teacher chooses the inputs and reviews the output. If you want to put that permitted zone to work this semester, Jellypod is free for educators.