The Education Department Makes AI a Federal Grant Priority as Students Say AI Is Harming Their Critical Thinking

Framing

On April 13, 2026, the U.S. Department of Education published a final rule in the Federal Register that permanently embeds artificial intelligence into the competitive criteria for every major federal discretionary grant program, effective May 13. The rule does not merely encourage AI adoption. It awards competitive priority to districts proposing projects that expand AI literacy, integrate AI into teacher professional development, and use AI for early intervention and special education services. This is not guidance. It is a procurement architecture embedded in grant regulation. Districts that have not begun building AI governance infrastructure will now feel that absence in the form of diminished grant competitiveness.

The timing creates a structural tension that every district leader must name explicitly. The same administration that shuttered the Office of Educational Technology in 2025, eliminating the primary federal entity charged with supporting evidence-based edtech adoption, has now issued a final rule making AI adoption a funding priority. The agency that removed the guardrails is now accelerating the traffic. The rule was finalized over significant stakeholder opposition: commenters raised concerns about cognitive risk, privacy, and the absence of safety evidence. The department acknowledged those concerns and finalized anyway. For districts, this means the evidence gap is not narrowing. The obligation to navigate it is intensifying.

RAND's March 2026 student survey data sharpens this picture. The share of middle and high school students using AI for homework rose to 62%, a 14-point increase in seven months. But the more consequential finding is that 67% of those students believe AI is harming their critical thinking. Students are making the performance-learning tradeoff calculation that most district policies have not yet articulated. When students are ahead of their institutions in naming a governance problem, institutions are behind.

The international research community is simultaneously building the conceptual scaffolding district leaders need. A peer-reviewed Delphi study published April 1, 2026, reached international expert consensus on eight components of K-12 AI literacy. The OECD is incorporating similar frameworks into the 2029 PISA assessment design. These developments signal that AI literacy is transitioning from aspirational to accessible at a global scale. Districts building AI literacy programs now have access to validated frameworks, but access is only useful if governance structures exist to implement them. The federal grant rule, the student perception data, and the international framework development converge on a single institutional imperative: governance architecture must be built ahead of adoption, not retrofitted after it.

Top Research and Policy Signals

1. U.S. Department of Education Final Priority on Advancing AI in Education

Source type. Federal regulatory document, published in the Federal Register; not peer-reviewed.

Citation (APA). U.S. Department of Education. (2026, April 13). Final priority and definitions: Secretary's supplemental priority and definitions on advancing artificial intelligence in education. Federal Register, 91(70), 18774 to 18780. federalregister.gov

The rule establishes AI literacy integration, educator AI professional development, and AI-supported special education services as competitive priorities in the department's discretionary grant programs, effective May 13, 2026. Applicants proposing projects aligned to these priorities receive competitive scoring advantages. Grant-eligible uses include: expanding AI and computer science education in K-12, professional development for educators to integrate AI into subject areas, AI-driven personalized learning tools, and AI for early intervention in special education. The department received substantial stakeholder opposition (commenters warned of cognitive risks, privacy exposure, and a lack of safety evidence) and finalized the rule despite those concerns. The rule also establishes definitions of key AI terms that will shape how districts frame AI work in future grant applications.

Districts that have an AI governance architecture (policy frameworks, designated oversight, and documented literacy programming) will be positioned to align grant proposals with these criteria. Districts that do not will face a new form of competitive disadvantage in federal funding, beginning with the first grant cycle after May 13.

Leadership implications. Grant directors and curriculum leaders must audit current and upcoming federal applications to ensure alignment with the new AI priority criteria before the May 13 effective date. Superintendents should recognize this rule as a structural shift: AI governance capacity is now a grant-competitive asset. Districts without documented AI literacy programs, designated AI oversight roles, and articulated professional development structures are not just behind on governance; they are behind on resources.

2. More Students Use AI for Homework, and More Believe It Harms Critical Thinking

Source type. Research report, survey-based; RAND Corporation institutional research; not peer-reviewed.

Citation (APA). RAND Corporation. (2026, March). More students use AI for homework, and more believe it harms critical thinking skills: Selected findings from the American Youth Panel. RAND. rand.org

Among middle and high school students surveyed in December 2025, 62% reported using AI for homework, up from 48% in May 2025, a 14-point increase in seven months. ChatGPT was used by 53% of respondents; Google Gemini doubled its share in the same period. Chatbots were the most commonly used category (60%), followed by writing helpers such as Grammarly (21%). More consequentially, 67% of students endorsed the statement that AI use "harms critical thinking skills," a figure up more than 10 percentage points in ten months. Female students expressed greater concern than male students (75% vs. 59%). Only approximately one-third of students reported that their school had a schoolwide AI policy, and many, especially in secondary grades, reported that AI rules varied by individual teacher rather than by school or district.

Study context. The report is a survey instrument, not a causal study. Student perception of cognitive harm is not the same as measured cognitive harm. That distinction matters for research interpretation, but it does not diminish the governance signal. A supermajority of students are independently concluding that AI use involves a trade-off that their institutions have not named in their policies.

Leadership implications. The critical-thinking concern is not a minority view. It is the dominant position among students now using AI. District leaders who have not yet addressed the performance-learning tradeoff in their AI policies are now behind their own student populations in acknowledging it. Schoolwide AI policies, reported by only one-third of students as existing, are the foundational governance documents from which role-specific guidance for students and educators must flow. Absence of that document is not a neutral condition. It is a governance failure with measurable equity implications.

3. AI Literacy for K-12 Education: An International Delphi Study

Source type. Peer-reviewed journal article, Interactive Learning Environments (Taylor & Francis); published April 1, 2026.

Citation (APA). AI literacy for K-12 education: An international Delphi study. Interactive Learning Environments. tandfonline.com

Using three Delphi rounds with interdisciplinary and international expert stakeholders, this study reached consensus on eight core components of AI literacy for K-12 education, organized under foundational AI knowledge, critical perspectives, and societal and ethical dimensions. This is one of the first internationally validated conceptual structures for K-12 AI literacy competency. Its development aligns directly with OECD's February 2026 announcement that AI and media literacy will be integrated into the 2029 PISA assessment framework, signaling that AI literacy is transitioning from a locally optional program to a globally benchmarked outcome. The international scope of both the Delphi study and the PISA framework decision means that U.S. districts are now designing AI literacy programs that will be assessed against international standards within three years.

Study context. International expert panel; Delphi methodology (three rounds); published April 1, 2026, in a peer-reviewed journal.

Leadership implications. Districts building AI literacy curricula now have access to internationally validated conceptual frameworks rather than internally generated definitions. The eight-component structure provides a defensible foundation for scope-and-sequence decisions, educator role definitions, and student competency benchmarks. When PISA begins measuring AI literacy in 2029, districts without structured programs will face a new form of comparative accountability. Building to a validated framework now, rather than designing locally and retrofitting later, is the governance-efficient path.

4. Artificial Intelligence in K-12 Education: An Umbrella Review

Source type. Peer-reviewed journal article, Computers and Education: Artificial Intelligence (Elsevier); published 2025.

Citation (APA). Artificial intelligence in K-12 education: An umbrella review. Computers and Education: Artificial Intelligence. sciencedirect.com

This umbrella review synthesized 102 systematic reviews of AI in K-12 education, mapping the research corpus across eight application domains. AI applications most frequently discussed in the literature include intelligent tutoring systems (45% of studies), learning analytics (18%), and automated assessment (12%). The review finds that teacher preparation for AI (understanding how to use, evaluate, and critically govern AI tools) has received substantially less research attention than AI application development itself. Ethical risks, technological limitations, and systemic barriers are identified as the primary challenges across the corpus, with equity and inclusion as the most underresearched dimensions. The authors develop a framework for mapping AIEd territory and a rubric for assessing the quality of systematic reviews, revealing gaps in data management, analysis transparency, and extraction methodology across the existing literature.

Leadership implications. The research base is wide on what AI tools can do and thin on whether educators are equipped to govern them. This asymmetry has direct implications for professional development design. Districts investing in AI deployment without parallel investment in educator AI governance capacity are replicating the exact gap the research identifies. Governance preparation, not just technical familiarity, must be treated as infrastructure rather than elective training. Procurement decisions made without educator governance competency in place create accountability exposure that the research base cannot retroactively resolve.

5. A Systematic Review of Teachers' Professional Development Needs for AI Integration in K-12

Source type. Peer-reviewed journal article, Computers, MDPI; published 2026.

Citation (APA). Artificial intelligence in K-12 education: A systematic review of teachers' professional development needs for AI integration. Computers, 15(1), 49. mdpi.com

Synthesizing 43 empirical studies on teacher AI professional development in K-12, this systematic review finds that technical training alone is insufficient for effective AI integration. Successful adoption requires a combination of pedagogical knowledge, positive attitudinal orientation, organizational support, and continuous, not episodic, training structures. Teachers who are actively involved in co-designing AI tools and curricula demonstrate higher ownership and better alignment with classroom needs. A four-level, process-oriented professional development framework is proposed: awareness, literacy, integration, and transformation. The review documents that most current professional development programs for AI remain at the awareness or basic literacy level, well short of the integration and transformation competencies required for effective governance.

Study context. Systematic review of 43 empirical studies; international scope; K-12 and secondary education focus.

Leadership implications. One-time AI workshops and platform demonstrations do not constitute professional development in AI governance. The four-level framework proposed here maps directly onto district HR planning: it implies tiered competency levels for educators, differentiated by role and instructional context, not uniform training events. Districts that treat AI literacy as a single professional development day are not meeting the evidence standard, and, given the new federal grant priority on educator AI professional development, they may not be meeting the grant-competitive standard either.

Emerging Strategic Themes

Theme 1. Grant competition as governance driver. The U.S. Department of Education's April 2026 final rule transforms AI governance from a professional recommendation into a competitive funding criterion. Districts with documented AI literacy programs, designated oversight roles, and structured professional development will be positioned to compete for federal discretionary grants beginning in the 2026-27 cycle. Districts without this architecture will not merely be behind on governance; they will be behind on resources. The rule creates a financial incentive structure that no prior AI guidance document has produced.

Theme 2. Student awareness outpacing institutional policy. RAND's December 2025 data reveal a governance inversion: students are actively naming the cognitive trade-offs of AI use at a higher rate than school policies address. When 67% of students believe AI harms critical thinking and only one-third attend schools with a schoolwide AI policy, the demand for governance from students is rising, while institutional response lags. This pattern (stakeholder awareness preceding policy) historically precedes regulatory action. Districts that wait for external pressure before naming the tradeoff will find themselves responding rather than leading.

Theme 3. AI literacy as an internationally benchmarked outcome. The convergence of the peer-reviewed Delphi consensus framework and OECD's 2029 PISA design signals a structural transition: AI literacy is moving from a locally defined elective to a globally benchmarked competency. U.S. districts that treat AI literacy as a technology elective rather than a core literacy domain will face a new category of comparative accountability within three years. The window to build structured, framework-aligned AI literacy programs before PISA 2029 is open now, and narrowing.

Theme 4. Educator governance capacity as the missing infrastructure layer. Two independent peer-reviewed reviews, published within months of each other, reach convergent conclusions: the research corpus is rich in AI applications but thin on educator preparation to govern them. This asymmetry is not accidental: educator governance preparation has not been treated as a first-class infrastructure investment. The new federal grant priority on educator AI professional development creates an external forcing function, but the underlying need precedes the grant incentive. Districts that build genuine tiered competency frameworks for educator AI governance will be differentiated from those that treat professional development as a compliance checkbox.

Watch This Week

• Maryland HB 1057. Maryland HB 1057 is in final House deliberations. If enacted, it becomes the most structurally detailed state AI governance mandate for K-12 in the country, with designated AI coordinator roles, statewide tool certification, and required local policies.
• OECD 2029 PISA. The OECD 2029 PISA AI and media literacy framework is in active development. Track design decisions at oecd.org/pisa. The framework being built now will define the international standard against which U.S. districts' AI literacy programs will eventually be compared.
• FutureEd 2026 legislative tracker. 53 bills across 25 states are active in classroom instruction and AI regulation. Monitor committee schedules for floor votes in active sessions, particularly in states with late-session windows.
• RAND American Youth Panel. A follow-up survey is expected mid-2026. The current trajectory (62% use, 67% critical thinking concern) will either accelerate or signal stabilization in the next data release. Both have governance implications.
• Federal grant applications. Watch for early-mover districts updating federal grant applications to align with the new ED AI priority criteria. Their language choices and program design decisions will establish a practical precedent for grant-competitive AI governance framing.

Novo Executive Summary

The week of April 13, 2026, produced the most consequential federal AI education policy action since the current administration took office: a final rule embedding AI literacy and adoption into the competitive criteria for every major federal discretionary grant program, effective May 13. That rule lands in a landscape where 62% of middle and high school students already use AI for homework, 67% believe it harms their critical thinking, and most schools still lack a schoolwide policy. Internationally, the research community is converging on validated AI literacy frameworks, and the OECD is building them into the 2029 PISA assessment design, creating a global accountability structure that U.S. district leaders have approximately 3 years to prepare for. Two major peer-reviewed reviews published this quarter reach convergent conclusions: educator governance capacity is the least developed dimension of AI in education, even as deployment accelerates. District leaders who have treated AI governance as a future-planning item are now operating against both a federal funding deadline and an international assessment timeline simultaneously. Novo Innovative Pathways partners with district leadership teams to build the governance architecture (role definitions, policy frameworks, literacy scope and sequence, and procurement criteria) that converts AI adoption from a compliance exposure into a strategic and competitive capability.

What Was Not Found

No peer-reviewed causal studies exist examining the effect of federal grant priority rules on district AI governance outcomes. The April 13 ED final rule has no evidence base supporting its outcome claims. It was finalized despite stakeholder opposition citing the absence of safety and efficacy research, and despite the department's own acknowledgment of that opposition in the comment response. Districts are being incentivized to adopt AI at a grant-competitive level while the evidence base for what they are adopting remains unestablished.

No peer-reviewed studies examine AI literacy program outcomes in high-poverty K-12 districts, schools serving English language learners, or schools with high proportions of students with disabilities. The international Delphi study establishes consensus on AI literacy components but does not address whether those components are equitably accessible across school resource levels. Districts serving the highest-need populations are making AI literacy implementation decisions without research specific to their students.

No causal research exists on the cognitive effects of sustained AI homework assistance in K-12 settings. RAND's student perception data documents widespread concern about critical thinking, but perception data is not outcome data. The mechanism by which AI homework use affects learning transfer, long-term retention, or metacognitive development has not been rigorously studied using causal designs in U.S. K-12 populations. Districts cannot claim research support for policies addressing this tradeoff because the research does not yet exist.

No peer-reviewed studies examine the effects of specific AI governance structures (coordinator positions, procurement review boards, policy review cycles) on district AI adoption outcomes, equity metrics, or compliance rates. Districts are being encouraged to build governance architectures without evidence for which architecture designs produce better outcomes. The governance infrastructure being mandated by state legislation and incentivized by federal grants has not been evaluated as an intervention.

Questions District Leaders Are Asking

What did the Education Department's April 2026 AI rule do?

On April 13, 2026, the U.S. Department of Education published a final rule in the Federal Register establishing AI literacy integration, educator AI professional development, and AI-supported special education services as competitive priorities in its discretionary grant programs. Applicants proposing aligned projects receive competitive scoring advantages, and the rule also defines key AI terms that will shape grant applications.

When does the federal AI grant priority take effect?

May 13, 2026. The department finalized the rule despite stakeholder comments warning of cognitive risks, privacy exposure, and a lack of safety evidence, and it acknowledged that opposition in its comment response.

What did RAND find about students using AI for homework?

Among middle and high school students surveyed in December 2025, 62 percent reported using AI for homework, up from 48 percent in May 2025. Sixty-seven percent agreed that AI use harms critical thinking skills, with female students more concerned than male students (75 percent versus 59 percent). Only about one-third said their school had a schoolwide AI policy. It is a survey of perceptions, not a causal study.

What is the international Delphi study on K-12 AI literacy?

A peer-reviewed study in Interactive Learning Environments, published April 1, 2026, used three Delphi rounds with an international, interdisciplinary expert panel and reached consensus on eight core components of K-12 AI literacy, organized under foundational AI knowledge, critical perspectives, and societal and ethical dimensions.

What does research say about teacher professional development for AI?

A systematic review of 43 empirical studies in Computers (2026) found that technical training alone is insufficient. Effective integration requires pedagogical knowledge, positive attitudes, organizational support, and continuous training, and most current programs remain at the awareness or basic literacy level. The review proposes four levels: awareness, literacy, integration, and transformation.

What evidence is missing on AI in K-12 education?

No peer-reviewed causal studies examine the effect of federal grant priorities on district AI governance, no studies examine AI literacy outcomes in high-poverty districts or schools serving English learners or students with disabilities, no causal research covers the cognitive effects of sustained AI homework help in U.S. K-12 settings, and no studies evaluate which district AI governance structures produce better outcomes.

Sources

Governance and Policy

U.S. Department of Education. (2026, April 13). Final priority and definitions: Secretary's supplemental priority and definitions on advancing artificial intelligence in education. Federal Register, 91(70), 18774 to 18780. federalregister.gov

K-12 Dive. (2026, April 13). How the Education Department will prioritize AI in awarding grants. k12dive.com

MultiState. (2026, April 9). How states are regulating AI in education this legislative session. multistate.us

FutureEd. (2026). Legislative tracker: 2026 state AI in education bills. future-ed.org

Research, Peer-Reviewed

[Author names could not be confirmed from abstract access alone; flagged]. (2026). AI literacy for K-12 education: An international Delphi study. Interactive Learning Environments. tandfonline.com

[Author names could not be confirmed from abstract access alone; flagged]. (2025). Artificial intelligence in K-12 education: An umbrella review. Computers and Education: Artificial Intelligence. sciencedirect.com

[Author names could not be confirmed from abstract access alone; flagged]. (2026). Artificial intelligence in K-12 education: A systematic review of teachers' professional development needs for AI integration. Computers, 15(1), 49. mdpi.com

Research, Reports and Surveys (Not Peer-Reviewed)

RAND Corporation. (2026, March). More students use AI for homework, and more believe it harms critical thinking skills: Selected findings from the American Youth Panel. RAND. rand.org

RAND Corporation. (2025). AI use in schools is quickly increasing, but guidance lags behind: Findings from the RAND Survey Panels. RAND. rand.org

OECD. (2026). OECD digital education outlook 2026: How can AI help human beings learn and grow? OECD Publishing. oecd.org