For two years, adopting a district AI policy was a matter of judgment, something a thoughtful cabinet did when it found the time. In 2026 it became a matter of statute, with a date attached. If you lead an Ohio district, that date is July 1, and it is the nearest edge of a national wave you will not be able to wait out.
The scale is no longer anecdotal. State lawmakers introduced 134 bills addressing AI in education across 31 states this session, according to the government-relations firm MultiState. The bills cluster around three demands (data privacy, classroom-use restrictions, and AI literacy), and a growing number have left the “proposed” column for the statute books. The question facing your board is no longer whether to have a policy, but whether yours will be authored deliberately or assembled under deadline pressure from a template you did not vet.
What the new statutes actually compel
The laws now in force are more prescriptive than the early voluntary guidance, and they differ in ways that matter for how you write your own policy.
Ohio is the bluntest. House Bill 96 requires every traditional district, community school, and STEM school to adopt an AI-use policy by July 1, 2026. The state published a model policy on December 30, 2025 that districts may adopt or adapt; it addresses student and staff use, data privacy, academic integrity, and the evaluation of third-party tools. Adopting the model wholesale is permitted, but a model policy is a floor, not a fit.
Maryland’s AI Ready Schools Act, approved May 26, 2026, takes the more coordinated route. Once the state department of education issues its guidance, districts have a fixed window to adopt aligned policies, designate a non-instructional AI coordinator drawn from the central office, and procure tools that comply with state standards. The coordinator requirement is the operative detail: Maryland is telling systems that AI oversight is now a named job, not a committee’s spare-time concern.
Idaho’s Senate Bill 1227, signed by Gov. Brad Little on March 26 and effective July 1, requires districts and charters to adopt AI-use policies aligned to a state framework, and writes a principle directly into law: AI may assist instruction and administration but “cannot replace or eliminate a human teacher.” For a board worried about where automation stops, that sentence is worth borrowing.
Oklahoma’s Responsible Technology in Schools Act (SB1734) is the most useful template for the high-stakes questions. It mandates human-in-the-loop use, bars AI from being the primary basis for grading, discipline, or placement decisions, and preserves a parent’s right to opt a child out of student-facing AI tools without academic penalty. Districts there must have a written policy before the 2027-28 year.
Why the deadline is really about training, not technology
The temptation is to read these laws as paperwork and respond with paperwork. The data argues against that reading. A Center for Democracy & Technology survey found that 85 percent of teachers and 86 percent of students had already used AI during the 2024-25 school year, but only about half reported receiving any training or guidance about it from their school. Your staff and students are already operating the tools; what the statutes are really forcing is the supervision layer that should have accompanied adoption.
That gap is the actual exposure. An unwritten policy does not mean no AI in your buildings; it means undocumented AI, used without guardrails, with the liability still accruing to the district. The laws are simply moving that liability from implicit to explicit.
The cautionary tale worth citing to your board
When a board member proposes the simplest response (block it), the useful precedent is New York City. The nation’s largest district banned ChatGPT from its devices and networks in January 2023, then reversed the ban roughly five months later, with its chancellor conceding the prohibition had “overlooked the potential” of the tools and the reality that students would meet them regardless. The reactive ban cost credibility and bought nothing durable. A policy that anticipates use ages better than one that pretends it away.
The artifact: a policy checklist that satisfies the statutes
Whatever your state requires, a defensible 2026 AI policy answers seven questions in writing. Bring these to your next cabinet meeting:
- Permitted and prohibited uses: for students and for staff, named concretely, not by vibe.
- The human-in-the-loop line: which decisions (grading, discipline, placement, special-education determinations) may never rest primarily on AI output.
- Data and privacy: what student data may enter which tools, with FERPA and state privacy law cited.
- Vendor and procurement standards: the questions every AI tool must answer before it touches a classroom.
- The opt-out: how a family declines student-facing AI without academic penalty.
- The named owner: who, by title, is accountable for AI oversight (Maryland now requires this; adopt it regardless).
- Review cadence: when the policy is revisited, because this statutory landscape will not hold still.
A district that can answer those seven, in language a board can defend in public, has met both the letter of the new laws and the standard they are reaching for. The deadline is the easy part. The policy that survives the year after it is the work.