Within the span of simply two years, synthetic intelligence has moved from an rising curiosity to an operational actuality in American Ok–12 lecture rooms. College students are utilizing it to draft essays and construct interactive research apps. Academics are utilizing it to generate lesson plans, differentiate instruction and develop assignments. Directors are utilizing it to summarize knowledge, create chatbots and set up groups. And most college districts are scrambling to replace or create insurance policies that mirror the ever-changing world of AI.
However what does an AI coverage panorama appear to be? What number of districts have formal AI insurance policies in any respect? What do these insurance policies say? And what does the distribution of approaches reveal concerning the state of readiness, fairness and strategic considering in Ok–12 schooling?
To reply these questions, we constructed the AI Faculty Coverage Database by utilizing a five-level AI coverage continuum to code 122 districts and faculties throughout 38 states after which analyzed the outcomes. What we discovered is a snapshot of a area that’s neither panicking nor confidently main: it’s ready, watching and managing uncertainty one teacher-directed determination at a time.
What This Snapshot Tells Us
The dataset doesn’t seize each district/college in America; it’s a structured pattern. However the patterns it reveals are constant and interpretable. Taken collectively, they level in the identical path.
-
Most districts/faculties are reactive. The dominant posture is conditional permission, teacher-delegated decision-making and a wait-and-see perspective towards systemic AI integration.
-
Some districts/faculties are nonetheless attempting to cease the clock. Almost 30% are actively proscribing or prohibiting AI use, an strategy that’s more and more tough to maintain as scholar entry to AI instruments extends far past college networks.
-
Solely a small fraction is daring to steer. Fewer than one-third of districts/faculties have revealed insurance policies that mirror real strategic governance: formal frameworks, outlined authorised makes use of, employees improvement and equity-centered design.
-
Geography issues. Regional variation in coverage progressiveness is substantial and displays structural variations in state-level funding, board tradition and steerage infrastructure.
-
State steerage is important however not ample. Districts/faculties in states with official steerage are usually not reliably extra progressive than these with out. Legally mandated steerage seems to drive adoption extra reliably than advisory frameworks.
-
Insurance policies are written for college students, not for districts/faculties. The dominant framing of AI coverage as scholar conduct governance leaves important institutional questions on employees use, procurement, fairness and organizational technique largely unanswered.
As we interviewed districts/faculties over the previous yr, it has change into clear that extra consideration should be paid to adults and methods. Considered one of our findings is that insurance policies emphasize college students with a deal with dishonest — when, surely, a lot of the districts we’re working with this yr have had larger dangers introduced by grownup employees members misuse, together with sharing a PDF copy of the copyrighted curriculum with a private LLM account and violating copyright agreements; sharing IEP and analysis data with out deidentifying scholar knowledge. These missteps maintain actual long-term threat for districts. The flexibility of our schooling methods to combine AI responsibly is important to our means to uphold knowledge privateness for workers, college students and households alike.
The 5 Ranges of the AI Coverage Continuum
Degree 1 – Professional-Innovation/AI Inspired: Actively promotes AI as a studying software; embeds AI literacy into curriculum; proactively deploys AI instruments and gives open entry to college students and employees; emphasizes fairness and alternative in AI use
Degree 2 – Guided Integration: Has revealed a proper AI framework or multilayered steerage doc that defines authorised makes use of, builds AI literacy and addresses privateness and ethics; has structured AI onboarding for college students and employees
Degree 3 – Conditional/Trainer Directed: Permits AI use however solely when explicitly licensed by the trainer; focuses on quotation, educational integrity and limits on substitution; doesn’t have a districtwide AI curriculum; delegates choices on AI use to particular person academics
Degree 4 – Restrictive/Integrity Targeted: Tightly restricts AI use or solely permits it when trainer provides permission; solely permits use of vetted/authorised AI instruments; locations sturdy emphasis on dishonesty enforcement and penalties when evaluating AI use
Degree 5 – Prohibited/No Use: Explicitly prohibits AI use or treats it as plagiarism/educational dishonesty; doesn’t present a pathway for licensed scholar AI use; makes use of self-discipline or provides college students zeroes as penalties for AI use violations
The 122 organizations spanned public college districts, constitution faculties and particular person faculties throughout all 4 main U.S. census areas. We coded district/college insurance policies primarily based on publicly out there paperwork together with board of schooling insurance policies, acceptable use insurance policies, employees and scholar handbooks and official district/college web sites. We tracked the standing of state-level steerage utilizing the Heart on Inclusive Design and Digital Studying (CIDDL) as a reference.
Total, 3.3% of districts/faculties operated at Degree 1, 27.9% operated at Degree 2, 44.3% operated at Degree 3, 17.2% operated at Degree 4 and seven.4% operated at Degree 5.
Most Districts Are within the Cautious Center
Essentially the most hanging discovering from our dataset is the focus of districts/faculties that function at Degree 3. The concept that AI is permitted, however solely when the trainer says so sums up the coverage posture of almost half (44.3%) of the districts/faculties within the dataset.
Degree 3 insurance policies share a standard logic: they don’t ban AI outright, however they don’t take a strategic place on it both. As a substitute, they delegate the choice to particular person academics. College students might use AI solely when explicitly licensed and are required to attribute AI-generated content material as such. College students are prohibited from submitting AI-generated content material as their very own. The district/college units boundaries; academics set the principles inside these boundaries.
This strategy is comprehensible given the tempo of AI improvement. Nevertheless it carries actual dangers. When AI governance is delegated completely to particular person academics, fairness of entry turns into uneven. College students in lecture rooms with AI-forward academics get basically totally different academic experiences than friends in the identical constructing whose academics stay restrictive. District-level technique is changed by classroom-level improvisation. A Degree 3 coverage usually means the superintendents and directors have began AI governance work however haven’t but developed a real strategic framework.
Almost 1 in 4 Districts/Faculties Limit or Ban AI
Though Degree 3 is the most typical coverage posture, a considerable share of districts/faculties (almost 25%) has moved in the other way: 17.2% function at Degree 4 and one other 7.4% function at Degree 5.
Degree 4 districts/faculties are primarily involved with educational integrity, emphasizing detecting and prohibiting the submission of AI-generated content material. Degree 5 districts/faculties take it a step additional and concern categorical bans on AI instruments completely.
Districts/faculties that ban AI are usually not essentially improper, however they’re working with no long-term plan. Prohibition is a holding sample.
Restrictive insurance policies are usually not randomly distributed. They cluster in particular states and areas. Amongst states with no official steerage, districts/faculties are likely to have extra restrictive or variable coverage postures. Florida (with a mean degree of 4.5 throughout all districts/faculties within the dataset), Texas (common of 5.0), Indiana (common of three.6) and Wisconsin (common of three.6) have disproportionately excessive concentrations of restrictive-to-prohibitive approaches.
It’s price noting that restriction and prohibition are usually not inherently improper. Some districts/faculties have made deliberate, values-based choices to restrict AI for pedagogical causes. However the findings from our dataset counsel that in lots of circumstances, prohibition is a proxy for coverage uncertainty — a approach of shopping for time moderately than an intentional governance alternative.
Energetic AI Promotion Is Uncommon
Solely 3.3% of districts/faculties within the dataset function at Degree 1. These are districts/faculties that embed AI literacy into the curriculum, actively deploy instruments throughout your entire district/college, present college students and employees with open and supported entry and emphasize fairness of entry as a coverage design precept.
One other 27.9% function at Degree 2, which means they’ve revealed a proper AI framework with steerage on authorised makes use of, employees onboarding and privateness and ethics.
Which means that almost one-third of districts/faculties have moved past reactive governance into proactive technique. Nevertheless it additionally signifies that two-thirds are nonetheless working reactively — responding to AI moderately than main with it.
Regional Variation Factors to Structural Inequities
The dataset reveals significant regional variation in how districts/faculties strategy AI governance. The Midwest lags notably behind different areas, with a mean coverage degree of three.2, about half some extent extra restrictive than the Northeast (2.7). This sample holds even after accounting for state AI steerage: Midwestern states like Indiana, Wisconsin and Ohio have substantial proportions of Degree 4 and 5 districts/faculties. The Northeast and West present probably the most embracing, pushed partly by early-adopter districts/faculties in California, New Jersey, Oregon, Virginia and Washington. The South is break up: some states are producing a few of the most progressive district-level coverage language within the nation, whereas others are among the many most restrictive insurance policies we reviewed.
These regional patterns matter as a result of they aren’t random. They mirror structural variations in state-level coverage infrastructure, edtech funding capability, college board cultures and the diploma to which districts/faculties really feel supported by state-level steerage. Thus, a scholar’s relationship with AI at school is, partly, a operate of the place they reside.
State Steerage Issues, However Its Affect Is Uneven
Of the 122 districts/faculties within the dataset, 63% are in states with formally revealed AI steerage from the state division of schooling. Nevertheless, solely 15.6% explicitly reference state steerage in their very own coverage paperwork. This decoupling of state steerage from native coverage improvement is among the most necessary findings from the dataset.
States have accomplished vital work: as of early 2025, greater than 28 states had revealed AI steerage for Ok–12 faculties. However that steerage shouldn’t be reliably making its approach into district/college coverage. By and enormous, districts/faculties are writing their very own frameworks with out anchoring to the sources states have developed, or they aren’t writing frameworks in any respect.
The state steerage is there. The query is whether or not districts/faculties understand it exists, belief it, or have the capability to place it into observe.
When state steerage turns into a authorized mandate, district- and school-level coverage follows. When it stays advisory, districts/college adoption is inconsistent. States which have revealed steerage however haven’t mandated adoption ought to contemplate what extra helps, resembling mannequin insurance policies, technical help and tiered implementation instruments, would assist districts/faculties transfer from consciousness to motion.
Most Insurance policies Communicate to College students, Not Methods
Almost two-thirds (65%) % of coverage paperwork within the dataset checklist college students as the first viewers. Solely 22% have insurance policies explicitly addressed to each college students and employees. Only one group has a coverage targeted totally on employees.
This distribution reveals a spot: most AI coverage in Ok–12 schooling is framed as scholar conduct governance, not as institutional technique. The insurance policies reply the query of what college students can do with AI way more usually than they deal with how the district/college ought to use AI to raised serve college students.
Insurance policies that focus completely on scholar habits miss the bigger organizational dimensions: How are academics being educated to make use of AI responsibly? How is the district/college vetting AI instruments for knowledge privateness and algorithmic bias? How is the district/college defending employees and scholar knowledge? How is management desirous about the function of AI in particular schooling, multilingual learner help, or psychological well being providers? These questions require coverage that speaks to the entire group, not simply to scholar handbooks.
Suggestions for District/Faculty Leaders
Based mostly on our findings, we suggest the next to superintendents and district/college directors:
-
Deal with AI coverage as technique, not compliance. A coverage that tells college students after they can and can’t use generative AI is a place to begin, not a vacation spot. Efficient AI governance requires district- or school-level strategic considering: What AI instruments will we use? What use circumstances for AI will we prioritize and systematize? How will we vet instruments and their outputs? How will we practice our employees? How will we guarantee equitable entry? These questions belong in board-level coverage, not simply scholar handbooks. They necessitate ahead considering commitments that enable for secure innovation by adults.
-
Have interaction your state’s steerage sources. Solely 15.6% of districts/faculties on this dataset explicitly reference state-level steerage — a missed alternative. State schooling companies have invested vital sources in AI frameworks, mannequin insurance policies and procurement steerage. Districts/faculties that aren’t anchoring to those sources are doing extra work than obligatory.
-
Transfer prohibition to a transition plan. Districts/faculties at present at Degree 4 or 5 ought to actually ask themselves if prohibition working. College students with smartphones have entry to AI across the clock. A coverage that prohibits AI on college networks with out addressing the broader context of scholars’ AI entry outdoors of college shouldn’t be a governance resolution. It’s a governance deferral. The query shouldn’t be whether or not college students will use AI however whether or not faculties will assist them be taught to make use of it properly.
-
Design for entry from the beginning. The Degree 1 districts/faculties on this dataset acknowledge that each coverage determination both expands or limits college students’ alternatives to profit from AI. Quite than treating fairness as a separate initiative, they construct concerns resembling system availability, web entry, accessibility, multilingual helps, skilled studying and authorised AI instruments into the coverage from the outset. As districts/faculties develop or revise their AI insurance policies, they need to ask explicitly whether or not every coverage determination will create or cut back disparities in scholar and/or employees entry to significant AI use. The reply ought to frequently form the coverage.
-
Construct coverage along with your group. Essentially the most sturdy AI insurance policies on this dataset emerged from multistakeholder processes: committees of academics, college students, dad and mom and directors working collectively. Coverage constructed with the group carries legitimacy that top-down mandates can not. It additionally surfaces sensible issues about educational integrity, knowledge privateness and scholar well-being that profit from various views.
What We Discovered
The AI coverage panorama in Ok–12 schooling displays a area in real transition. The governance dialog has clearly begun. However the distribution of that progress suggests a area that has not but discovered its footing.
What the information clarify is {that a} single coverage posture shouldn’t be the trail ahead, neither is a single software or use case. What districts/faculties want is a sustained governance course of — one which strikes from reactive administration of scholar habits to proactive design of AI-enabled methods, connects native observe to state-level sources and treats entry and alternative as a non-negotiable design precept.
The districts/faculties at the forefront of this dataset provide a proof of idea. They aren’t ready for good readability about the place AI is headed. They’re constructing frameworks resilient sufficient to evolve, clear sufficient to maintain group belief, considerate sufficient to supply security and safety and versatile sufficient to make sure that the usage of AI by adults can help outcomes for all college students.
That could be the mannequin price scaling.
