The K-12 Engine of American Innovation: A National Strategy for Education in the Age of AI

Executive Summary
 
As Congress considers various legislative frameworks to establish a national posture on artificial intelligence (AI), many important and timely issues are front and center. Designing strong governance, reinforcing research infrastructure, and anticipating labor market realignments are all critical parts of establishing a strong foundation for the broader society and economy. However, current debates often overlook a critical foundation of America’s future competitiveness—our K-12 education system.

Frontier laboratories have observed how AI is actively accelerating its own capabilities, sparking broader discussion around recursive self-improvement. Yet educational institutions need not await futuristic milestones to feel this rapid momentum. Classrooms across the country are facing immediate adoption, as students leverage advanced tools long before local leaders can establish coherent governance or assess instructional value. 

A passive posture toward empowering learners presents profound risks. Cultivating deep AI literacy, critical thinking, and student agency must be recognized as both an urgent pedagogical commitment and an economic imperative. Consequently, federal leaders must equip state and local education agencies with technical assistance, rigorous evidence frameworks, professional development, and sustainable funding, guaranteeing that innovation remains anchored in community needs and student safety.

To that end, this Big Bets policy memo proposes that any comprehensive federal AI legislation include a five-pillar national strategy for K-12 education, including: Cohesive Support, Responsible Innovation, Skill-Building Pathways, System Capacity, and School Models. Rather than relying on the unchecked deployment of commercial technologies, this approach centers pedagogy and student safety. By establishing national technical assistance, regional evidence sandboxes, localized AI literacy pathways, and funding modernizations, we can build the infrastructure needed for responsible innovation and implementation, while improving student outcomes and increasing access for communities.
 
Problem & Opportunity
 
Unregulated K-12 Adoption: America’s nearly 100,000 public schools are currently unregulated testing grounds for rapidly accelerating student AI adoption. As of June 2026, more than four out of five teens had used AI tools, with more than two-thirds reporting regular use.
 
Yet, institutional readiness is dangerously lagging. There is a glaring disconnect between the presence of AI and school systems' ability to oversee it, with only 18 percent of U.S. principals reporting having rules around AI use—a number that drops to 13 percent in high-poverty schools. Schools face compounding challenges, including a dearth of technical expertise, professional learning, evidence bases, and procurement support.
 
Without federal support and coordination, we risk exacerbating systemic inequities, creating a divided future where affluent districts harness AI for pedagogically-sound enhancements while under-resourced schools become passive consumers exposed to unsafe tools and use cases. 
 
Limits of Market Forces: We cannot rely solely on market forces to solve this problem. The industry's "move fast and break things" approach to AI will not work when children's developmental needs are at stake. Commercial vendors are heavily incentivized by user acquisition, engagement metrics, and data aggregation—rather than by rigorous evidence of instructional value.
 
This unchecked market push risks eroding student confidence and trust. According to recent data, 67 percent of students believe that the more they use AI for schoolwork, the more it will harm their critical thinking—a 10 percentage point increase in skepticism in under a year. While access is expanding, only 56 percent of students feel confident they will know how to use AI in their daily lives post graduation.
 
Furthermore, while public-private partnerships can drive innovation, schools remain vulnerable if commercial providers dictate the terms. These partnerships require consistent, rigorous expectations to ensure student safety and proven educational outcomes. We must shift from a model where technology is simply delivered to communities, toward one where innovation is co-designed with communities. 
 
Economic Imperative: The economic stakes are significant and our global competitors are not waiting. Nations like South Korea, China, and Singapore have already begun embedding AI competencies into their curricula. With the skills needed for many jobs expected to shift by up to 65 percent by 2030, lacking a coordinated strategy risks squandering public funds on fragmented, ineffective local software contracts.
 
Perhaps more importantly, an American workforce lacking competitive AI literacy will severely handicap our global standing. By embedding an education framework into our national AI strategy, Congress can transform K-12 education from under-resourced improvisations into a proactive engine of economic resilience.
 
Navigating Screen Time: This shift toward AI arrives amid intense, justified scrutiny over student screen time and a growing “tech-lash” fueled by data privacy anxieties. The U.S. Department of Education has addressed this tension directly, cautioning that recreational and educational technology should not be treated as synonymous, and that the latter should be evaluated on instructional value rather than screen time alone.
 
A coordinated national strategy provides a framework for a pragmatic, balanced approach that addresses these vulnerabilities directly. It moves the conversation beyond binary policy options—total bans versus unregulated access—by establishing developmentally appropriate guidelines for high-quality, intentional AI use that complements human instruction and interaction.
 
Proposed Solution
 
As K-12 education designed to serve all learners is foundational to American competitiveness in the AI era, the federal government must move beyond fragmented policies and enact a unified, proactive framework. We propose integrating the following five pillars into any comprehensive federal AI legislation, grounding implementation in four shared priorities—educational benefit, privacy and safety, educator and system capacity, and access and opportunity.
 
These five interrelated pillars form an integrated framework to comprehensively resolve these challenges. National technical assistance infrastructure establishes immediate guidance for local decisions, complementarily paired with regional evidence sandboxes to rigorously validate safety and pedagogical effectiveness. Parallel investments in localized AI literacy pathways ensure age-appropriate skill-building, while modernized block grants supply the operational capacity, governance, and infrastructure necessary for long-term execution. Finally, targeted innovation guardrails empower communities to co-design transformative school models without sacrificing student well-being.

Pillar 1: Cohesive Support – Artificial Intelligence Technical Assistance Center 
 
School districts are navigating a complex, unmapped landscape of AI tools. Many districts lack the expertise to safely procure and effectively integrate AI, leading to disparities in local policy and opening the door to massive pedagogical and security risks. To build capacity, Congress should fund national technical assistance infrastructure to support State and Local Educational Agencies (SEAs and LEAs).
 
While this infrastructure could be established as an entirely new federal hub—such as an Artificial Intelligence Technical Assistance Center (AITAC) modeled on the U.S. Department of Education’s Privacy Technical Assistance Center (PTAC)—Congress could also efficiently deliver this support by embedding dedicated AI capacity-building mandates within existing networks like the Comprehensive Centers or Regional Educational Laboratories. 
 
Regardless of the exact vehicle, this infrastructure must drive educational benefit and privacy and safety by facilitating evidence sharing, implementation guidance, and procurement recommendations. To fulfill its mission, AITAC must also develop public-facing resources to help schools meaningfully engage communities, such as parent communication toolkits, student responsible use frameworks, and transparency guidelines to build local trust.
 
To ensure access and opportunity, this national infrastructure should deploy a cohort of competitively compensated teacher leaders—modeled after Lead For America’s American Connection Corps—to assist under-resourced districts in building coherent AI instructional visions and localized strategies. Providing strong professional and financial incentives for these teacher leaders is essential for a sustainable talent pipeline. Furthermore, while private initiatives like Anthropic’s "Claude Corps" have been launched to offer similar support for the nonprofit sector, a well-resourced public option for K-12 ensures the federal government maintains a strict separation between corporate influence and educational decision-making.
 
Pillar 2: Responsible Innovation – Regional AI Evidence Sandboxes
 
Schools face an unprecedented number of choices for AI-powered instructional tools, yet lack a basis for judging their effectiveness. Evidence generation has not kept pace with the rate of product development, resulting in a lack of information for schools and districts.
 
Therefore, the federal government should take on the burden of establishing Regional AI Evidence Sandboxes, designed not merely as a network of testbeds, but as knowledge-generation infrastructure to identify how AI can drive instructional value and measurable student outcomes. These regional sandboxes could be built on the framework of Utah’s Senate Bill 322, which proposes pilot trials with voluntary participation from schools and strict oversight by governing bodies. To ensure privacy and safety, entry into these sandboxes must operate under modernized guidance that clarifies data privacy obligations under relevant federal legislation. 
 
These environments must also mandate proactive risk mitigation, including data collection and use limits, red teaming for accuracy and bias, and "kill switches” that immediately pause or terminate a pilot upon breaching pre-defined risk thresholds, detecting severe algorithmic bias, or identifying data privacy violations during continuous monitoring. 
 
To build educator and system capacity, these sandboxes should prioritize deep practitioner and school community involvement in the design of research, testing of AI use cases, and interpretation of results. Finally, to place access and opportunity at the forefront, these environments must be architected for scalability. They should be required to intentionally examine impact across diverse student populations, demographic contexts, and varying levels of socioeconomic status and digital connectivity. 
 
Rather than acting as a mandatory pre-adoption gatekeeper, these sandboxes should be designed to generate rigorous evidence that schools and districts can voluntarily draw upon to inform their own local procurement decisions, ensuring tools demonstrate clear educational benefit against established standards before broad deployment.
 
Pillar 3: Skill-Building Pathways – AI Literacy Opportunities 

The U.S. is falling behind international competitors like China and the European Union in formalizing an instructional progression for AI literacy. To close this gap, Congress must provide robust resources to assist the development of AI literacy in a coherent, locally-driven manner. Through vehicles such as new competitive grants, the federal government should act as a catalyst and support states and districts as they integrate AI concepts into existing standards and instructional resources.

While avoiding top-down mandates, these grants should advance:
 

• Educational benefit by prioritizing judgment, critical thinking, and evaluation of AI outputs; 

• Privacy and safety by embedding media literacy and responsible use principles; 

• Access and opportunity by ensuring competencies are integrated across core subjects for all students, rather than siloed exclusively in advanced computer science classes; and

• Educator capacity by ensuring teachers receive the training, planning time, and ongoing support needed to implement this content confidently and effectively.

Emerging policy measures, such as the proposed American Leadership in AI Act, represent a constructive starting point by directing National Science Foundation investments tailored for underserved rural, Tribal, and low-income populations. Federal leaders must build upon this momentum by deliberately expanding safeguards for learners with disabilities and ensuring these research insights translate into actionable, high-impact implementation frameworks for local school systems.

Furthermore, the program must move beyond abstract research to ensure direct educational benefit by mandating the translation of the U.S. Department of Labor’s AI Literacy Framework into ready-to-use, age-appropriate K-12 resources, preventing the wasted time and resources of creating entirely new frameworks from scratch. Finally, implementation should be guided by an advisory council of educators, parents and caregivers, developmental researchers, industry leaders, and disability advocates to direct investments and measure impact.
Pillar 4: System Capacity – Modernized EETT Program
 
Many SEAs and LEAs lack the sustainable operational funding required to execute AI integrations. There is a critical need for structured support in leadership, procurement, planning, evaluation, and instructional vision. We propose revitalizing Title II-D of the former No Child Left Behind Act—the Enhancing Education Through Technology (EETT) program—into “EETT 2.0,” a modernized federal block grant focused on sustainable AI integration.
 
Similar to its predecessor, EETT 2.0 would require SEAs to submit comprehensive AI implementation plans to secure grant funding, which may also be incorporated into existing state educational technology strategies. To ensure these plans are actionable, these grants should support both foundational planning—such as hiring state AI coordinators and contracting for technical assistance—and full-scale implementation. Specifically, state plans must explicitly address:
 

• Instructional Vision: Plans must detail how SEAs will equip educators to use AI to enhance teaching, ensuring the technology supports practice and drives direct educational benefit.

• Safeguards: Plans must outline rigorous procurement processes and governance to establish privacy and safety as foundational elements of all technology adoption.

• Evidence Generation: Plans must establish mechanisms to evaluate AI tools and measure the return on investment of implementations on student learning.

 
The majority of EETT 2.0 funds must be distributed via subgrants directly to districts for local planning, targeted AI professional development, community engagement, and reinforcing AI-ready technological infrastructure. This guarantees that a district's baseline capacity to operate safely in the AI era is secured regardless of its local tax base, a win for access and opportunity. Congress should also explicitly encourage the use of existing Titles II-A and IV-A authorizations under the Every Student Succeeds Act to supplement these efforts, providing a flexible funding portfolio.
 
Pillar 5: School Models – Guardrails for Innovation Grants
 
The public education system must build the capacity to tackle persistent problems of practice by piloting highly innovative, AI-integrated school models. But these experiments must never compromise student safety, data privacy, or fair access to high-quality instruction. As education expert Michael Horn notes, the ultimate goal must be to fundamentally rethink and redesign school models to optimize for educational benefit, rather than simply layering new AI tools on top of an outdated, industrial-era delivery system.
 
To achieve this, the federal government must incentivize responsible experimentation with new, AI-integrated models that generate actionable insights. The federal Charter Schools Program offers one vehicle for testing such models across diverse settings, although it need not be the only mechanism used to foster this innovation.
 
To ensure privacy and safety, all public entities receiving such innovation funds must be required to adopt strict AI guardrails, undergo rigorous algorithmic impact assessments, and fiercely protect student information. Importantly, these pilot models must be community-informed—requiring deep, sustained engagement with parents, students, and educators to co-design school models that reflect the unique needs, values, and aspirations of their localities.
 
Finally, to prioritize access and opportunity, these models must be scalable and tested across diverse demographics, establishing mechanisms for widespread dissemination of best practices to build broader educator and system capacity.
 
Conclusion
 
Comprehensive federal AI legislation represents a vital policy milestone, but its ultimate success hinges entirely on preparing the next generation to oversee, operate, and innovate the technologies that the policy aims to govern. America's future depends on sustaining those who can implement technologies responsibly and equitably. Congress should prioritize K-12 education to develop the critical thinking, literacy, and human judgment necessary to responsibly navigate an AI-augmented world.
 
Ultimately, K-12 education is the critical "social layer" of America's AI infrastructure. By investing boldly in Cohesive Support, Responsible Innovation, Skill-Building Pathways, System Capacity, and School Models, the U.S. can build a unified framework that secures our competitive edge, ensures broad-based prosperity for students of all backgrounds, and places human agency squarely at the center of the classroom.