States are taking different approaches to setting policies and guardrails around how artificial intelligence should be used responsibly. Governors have launched task forces. State education agencies have established committees and issued non-binding guidance. Legislatures have passed laws requiring districts to establish AI policies.
Our research shows that of 29 state education agency respondents to a November 2025 survey, 93% said their state or territory had communicated or planned to communicate a vision for AI in K-12 education. Yet only 43% said they had assisted or planned to assist districts with implementation support, like procurement or vendor pricing. And among the 39 states and territories we studied for our state early adopter database, just 11% had reached what we call the “integrator” stage: moving beyond vision and guidance to pilot programs, evaluation, procurement assistance, standards revisions and implementation strategies.
Get stories like this delivered straight to your inbox. Sign up for The 74 Newsletter
AI presents an unusual policy challenge because both the technology and public opinion continue to evolve quickly. Guidance and task forces alone are not enough. School and district leaders need practical support from states to help them navigate implementation challenges, such as vetting products, generating evidence on what works and building awareness and trust in how AI is used. And states need infrastructure that connects policy to practice and turns scattered local experimentation into shared knowledge. Many state education agencies right now lack the funding or internal expertise to lead all these efforts.
Given these challenges, intermediary organizations play an important, and increasingly imperative, role in AI integration across schools. A majority of participants in our study reported pursuing partnerships to help with strategy development, implementation or both. Respondents told us these partnerships help expand state capacity and reach, particularly for those with limited internal staff and AI expertise.
To better understand what these partnerships can look like, we focused on Tennessee’s State Collaborative on Reforming Education, or SCORE. SCORE is a statewide nonprofit with no formal authority over district decisions. Its influence comes instead from longstanding relationships with state leaders, districts, colleges, education funders and philanthropies, and technology providers. Those relationships allow SCORE to play roles that are difficult for many state agencies to sustain on their own.
For example, SCORE helps Tennessee leaders make sense of a rapidly changing landscape by convening local stakeholders and translating their experiences to inform state strategy. After Tennessee required districts to adopt AI policies addressing safety, academic integrity and risk mitigation, SCORE convened a working group of K-12 and higher education leaders, researchers, industry representatives and other state partners. The group met regularly for roughly 18 months, not merely to trade updates, but to compare strategies, surface implementation problems and ask where AI could support state goals for literacy, instructional quality and career readiness.
From these conversations, SCORE drafted recommendations for Tennessee’s AI Advisory Council that were ultimately reflected in the state’s Action Plan. It also used lessons from districts to recommend ways of integrating AI literacy skills into the state’s English language arts standards; the state Board of Education considered those recommendations during its review process.
SCORE also helps districts navigate an increasingly crowded marketplace of products. Its members spend significant time evaluating emerging tools, understanding district needs and sharing what they learn with educators. SCORE does not endorse particular tools or programs, or dictate purchasing decisions. But in a market where new products appear almost daily, that kind of trusted curation can reduce uncertainty without replacing local decision-making.
Coursemojo, an AI-enabled literacy tool, illustrates how this can work. SCORE studied the product because early evidence suggested it could embed additional reading support for struggling students within core middle school instruction. After a 2024-25 pilot in Sumner County Schools, SCORE partnered with a local foundation to help fund a larger cohort: seven districts, 20 schools,and more than 4,000 students in 2025-26.
Teachers reported using the tool’s insights to adjust instruction, while students engaged more deeply with texts and revised their work in response to feedback. All participating district leaders reported being highly satisfied, and teachers said the tool had positively affected student learning. Districts subsequently renewed and expanded the work, often with their own funds. Those findings are not a final verdict on the product, but they provide important evidence about promising tools, the conditions that need to be in place for them to make a difference and the feasibility of implementing them across the district.
The Tennessee SCORE model will not look the same in every state, nor should it. An intermediary assisting state AI implementation efforts might be a nonprofit, a university, a technical service provider or a research organization. What matters is that the organization has trust, staying power and the ability to connect people who do not routinely learn and work together.
Without that coordinated support, better-resourced districts may be able to experiment while smaller and higher-need systems may struggle to evaluate products, establish safeguards or provide meaningful professional learning. In this way, states risk repeating a familiar pattern: widened inequities in students’ access to high-quality learning experiences and appropriate protections.
The next phase of state AI leadership will be defined less by who publishes the next guidance document than by who builds the infrastructure for continuous learning. Policies establish direction. Partnerships can expand capacity to pursue that direction. Together, they offer states their best chance to move from isolated experiments toward coherent, responsible AI integration that benefits every school and student.