AI Policy Has Moved From Guidance to Governance
A visual 50-state scan of K-12 AI policy, showing how states are moving from voluntary guidance to enforceable district governance.
Explore the 50-state AI policy landscape.
Filter by state, region, or policy posture. Each tile shows a state abbreviation and implementation signal score. Click any state for the governance readout.
Color shows policy posture. Number shows implementation signal score from the combined scan. Source trackers are current-state indicators, not legal advice. Beige cardigan successfully avoided.
Executive take
K-12 AI policy has crossed the line from guidance into governance.
The first wave asked whether schools should use AI. The current wave asks who is accountable when they do: who approves tools, who sees student data, who explains model use to parents, who trains teachers, and who keeps a human in the decision loop.
The combined scan found 37 states with published statewide K-12 AI guidance, 3 more in formal development, and a growing set of enacted laws requiring districts to adopt aligned AI policies, embed AI literacy, or restrict high-risk uses.
That means the smart district move is not to wait for a perfect state framework. It is to build a local governance loop now.
AI policy is no longer a technology question. It is a trust architecture question.
The national picture
| Signal | Count | What it means |
|---|---|---|
| Published statewide guidance | 37 | Statewide K-12 AI guidance, toolkit, framework, or resource hub identified |
| Developing or legislated | 3 | Task force, enacted framework mandate, or formal working draft without final statewide framework |
| Active bills or district signal | 5 | Meaningful legislative or major district activity, but no finalized statewide guidance |
| No statewide guidance identified | 5 | No finalized statewide K-12 AI guidance in the source set |
Two counts matter, and they should not be mixed up:
- Guidance count: the statewide education guidance landscape. The live-source scan found 37 states with published K-12 AI guidance or official resource hubs.
- Law count: the enforceable governance landscape. FutureEd tracks 71 AI-in-education bills across 27 states in 2026, with 11 enacted 2026 education-focused bills in the source set. The uploaded report counts 13 binding-law states when broader and earlier AI laws are included.
Both are useful. They are not the same thing. Policy work gets sloppy fast when those definitions blur.
The pattern underneath the noise
Across the states, the same five building blocks keep showing up:
- Human-centered use. AI should support educators, not replace them.
- Data privacy and vendor limits. FERPA, COPPA, data minimization, and restrictions on using student data to train models.
- AI literacy. Students need judgment, not just prompt tricks.
- District policy. State guidance is turning into board-adopted local requirements.
- Evidence loops. Pilots, feedback, incident reporting, and outcome checks are still weak, but they are where the serious states are heading.
The policy center of gravity is moving from “Can teachers use ChatGPT?” to “What system of adult judgment surrounds every AI use?”
Highest maturity signals
This maturity score is not a legal ranking. It is an implementation signal: published guidance, enacted law, district policy requirements, literacy/curriculum activity, privacy guardrails, and evaluation/procurement signals.
| State | Posture | Signal score | Binding-law note from uploaded report |
|---|---|---|---|
| Maryland | Published guidance | 8 | Yes |
| Oklahoma | Published guidance | 8 | Yes |
| Virginia | Published guidance | 8 | Yes |
| Alabama | Published guidance | 7 | Yes (HB 329) |
| Washington | Published guidance | 7 | Partial |
| California | Published guidance | 6 | Partial |
| Ohio | Published guidance | 6 | Yes |
| Utah | Published guidance | 6 | Yes |
| Idaho | Developing / legislated | 5 | Yes |
| Mississippi | Published guidance | 5 | Partial |
Watchlist states for district leaders
| State | Current posture | Why it matters |
|---|---|---|
| Kansas | No statewide K-12 AI guidance identified | No KSDE statewide framework identified. USD 259 already has Board Policy P1231, which makes WPS an early local mover. |
| Maryland | Published statewide guidance | Most complete operating model: AI coordinators, local policies, PD, collaborative, and tool certification direction. |
| Oklahoma | Published statewide guidance | Human-in-loop, parent disclosure, opt-out rights, and no AI as primary basis for high-stakes student decisions. |
| Tennessee | Published statewide guidance | Early district-policy mandate plus mental-health AI restrictions and a family private-right-of-action signal. |
| South Carolina | No statewide guidance; active bills or major district policy | Pending bills point toward strict opt-in consent, data minimization, and teacher-replacement prohibitions. |
| Idaho | Developing / legislated | Enacted framework requirement, but the actual state framework still matters. Watch the details, not just the headline. |
| Georgia | Published statewide guidance | Traffic-light classroom model plus reported SB 179 district implementation requirements. |
| New York | No statewide guidance; active bills or major district policy | No statewide P-12 guidance in the live scan, but NYCPS and broader model-developer law make it impossible to ignore. |
| Rhode Island | Published statewide guidance | Source conflict: live scan found RIDE LEA guidance; uploaded report says local control. Verify directly before citing. |
| Wyoming | Published statewide guidance | Source conflict: live scan found policy-development guidance; uploaded report says no statewide guidance. Verify before citing. |
Kansas and Wichita: the useful tension
Kansas has not published a statewide KSDE AI framework in the combined source set. That places Kansas in the minority of states where districts are still largely operating without a state-level anchor.
But Wichita Public Schools is not starting from zero.
USD 259 adopted Board Policy P1231 on Artificial Intelligence in 2025. That matters. In a no-guidance state, a board-adopted AI policy becomes more than a compliance document. It becomes operating infrastructure.
The gap is that local policy alone does not solve the hard parts:
- Vendor vetting: districts need model contract language, data-use restrictions, bias/accuracy review, and auditable vendor claims.
- Parent transparency: states are moving toward notification, opt-out, and consent expectations.
- Professional development: teachers cannot implement responsible AI policy by vibes and a PDF.
- Student literacy: AI literacy needs to become part of instruction, not just a staff handbook paragraph.
- Monitoring and surveillance: the Lawrence Public Schools v. Gaggle Fourth Amendment case is a real Kansas watch item for districts using AI monitoring tools.
WPS is ahead of many Kansas peers because P1231 exists. The next move is implementation architecture.
What a district should build now
A practical district AI governance loop has five parts:
| Layer | District question | Concrete output |
|---|---|---|
| Policy floor | What is allowed, prohibited, disclosed, and reviewed? | Board policy plus staff/student acceptable-use guidance |
| Procurement gate | Which tools can touch student data? | Vendor rubric, contract clauses, approved-tool process |
| Learning design | How does assessment change when AI exists? | Classroom guidance, citation norms, defense/process tasks |
| Capacity | Who knows how to use AI responsibly? | Role-based PD for teachers, leaders, support staff, and students |
| Evidence loop | Is this helping learning, work, trust, or safety? | Pilot metrics, incident log, review cadence, public-facing updates |
The districts that get this right will not be the ones with the longest AI policy. They will be the ones with the clearest operating loop.
What state leaders are still missing
Even the better state frameworks tend to underbuild the same things:
- Vendor transparency. Few require auditable interaction records, bias/accuracy evidence, or clear model-training disclosures.
- Evaluation infrastructure. Districts are told to be careful, then left to invent evaluation from scratch.
- PD funding. Guidance without funded teacher learning is mostly decorative.
- Parent communication. Trust is becoming a formal policy requirement, not a nice-to-have.
- High-stakes decision rules. States are starting to prohibit AI as the primary basis for grading, discipline, placement, mental-health screening, and similar decisions. This should be universal.
The line I would use with a board
If I had to compress the whole report into one board-facing sentence:
We are not adopting AI because it is new. We are governing it because it is already here.
That is the posture. Not panic. Not hype. Governance.
Sources and scope
This post combines a live-source scan with an uploaded 50-state DOCX report prepared for WPS context. Core sources included TeachAI, Playlab, AI for Education, FutureEd, ExcelinEd, Digital Promise, Education Commission of the States, Code.org Advocacy Coalition, MultiState, and state education department documents.
Download the visual PDF report here: K-12 AI Policy 50-State Report