The Training Gap: How 134 State Bills, a $23 Million Union Academy, and a Federal Mandate Are Racing to Catch Up With AI in the Teacher’s Classroom
Eighty-five percent of teachers used AI last year, but fewer than half received any training. Now a federal order, a union-industry partnership, and a wave of state laws are trying to close the gap.
The Scale of the Problem
Artificial intelligence has entered K-12 classrooms faster than any prior educational technology. According to survey research from the Center for Democracy & Technology (CDT), 85 percent of teachers and 86 percent of students reported using AI during the 2024-25 school year[1]. Yet less than half of teachers and students received any training or information about AI from someone at their school[1]. A separate RAND Corporation study of nationally representative school districts found that roughly half of districts had provided any teacher training on generative AI as of fall 2024 — double the proportion from the previous year, but still leaving the other half of districts with no training infrastructure at all[2]. According to RAND researchers, the teachers who did receive training typically described it as a “one-off” event rather than ongoing professional development[2].
The problem is not merely quantitative but qualitative. CDT’s research found that only 11 percent of teachers say they have been trained on how to respond if they suspect a student’s AI use is detrimental to their well-being — such as hurting self-esteem or encouraging risky behavior[1]. Seventy percent of teachers expressed concern that students’ use of AI was preventing them from learning important skills[1]. The gap between adoption and preparedness is precisely the vacuum that state and federal policymakers are now rushing to fill.
The Federal Framework: Executive Order 14277
On April 23, 2025, President Trump signed Executive Order 14277, “Advancing Artificial Intelligence Education for American Youth,” which established the White House Task Force on Artificial Intelligence Education and made AI literacy and educator training explicit federal policy[3]. The order declares it U.S. policy “to promote AI literacy and proficiency among Americans by promoting the appropriate integration of AI into education, providing comprehensive AI training for educators, and fostering early exposure to AI concepts and technology to develop an AI-ready workforce”[3].
The order is specific about teacher training. Section 7 directs the Secretary of Education, within 120 days, to prioritize the use of AI in discretionary grant programs for teacher training authorized under the Elementary and Secondary Education Act and Title II of the Higher Education Act[3]. The four enumerated priorities reveal a clear theory of change:
- Reducing time-intensive administrative tasks — The logic is straightforward: if AI can cut the hours teachers spend on paperwork, grading drafts, and lesson-plan formatting, it frees them for the human work that machines cannot do. This frames AI not as a replacement for teachers but as a labor-saving tool that protects the teacher-student relationship by reclaiming time.
- Improving teacher training and evaluation — AI can analyze classroom patterns and provide feedback to teachers at scale, potentially making professional development more data-driven and individualized than traditional observation models allow.
- Professional development for all educators to integrate AI fundamentals into all subject areas — Rather than treating AI as a discrete computer science topic, this priority recognizes that a history teacher using AI-assisted research tools or an art teacher exploring generative imagery needs AI fluency just as much as a coding instructor. The principle is that AI literacy is now a cross-curricular teaching competency.
- Professional development in foundational computer science and AI for standalone courses — This ensures that teachers in dedicated CS and AI courses have the depth of knowledge to teach these subjects as disciplines, not just as tool-use tutorials.
The Department of Education’s Grant Guidance
Three months after the executive order, on July 22, 2025, the U.S. Department of Education issued a Dear Colleague Letter (DCL) outlining how school districts may use existing federal formula and discretionary grant funds to support AI integration[4]. The guidance identified three categories of allowable AI use: AI-based high-quality instructional materials, AI-enhanced high-impact tutoring, and AI for college and career pathway exploration and navigation[4].
Secretary of Education Linda McMahon framed the guidance as both an opportunity and a guardrail. “Artificial intelligence has the potential to revolutionize education and support improved outcomes for learners,” McMahon said, adding that the guidance “emphasizes the importance of parent and teacher engagement in guiding the ethical use of AI and using it as a tool to support individualized learning and advancement”[4]. The DCL also overviewed principles for responsible AI adoption, including attention to user privacy and the importance of teaching students about appropriate AI use in the context of social media[4].
Simultaneously, McMahon announced a proposed supplemental grantmaking priority — her fourth — that would “promote the appropriate integration of AI into education”[4]. The priority specifically calls for supporting professional development for educators on teaching AI and computer science fundamentals, and encourages the use of AI technologies to “enhance classroom efficiency, reduce administrative burdens, and improve teacher training and evaluation”[4]. The priority was published in the Federal Register for a 30-day public comment period ending August 20, 2025[4].
The Union-Industry Partnership: A $23 Million Academy
While the federal government was setting the framework, the American Federation of Teachers (AFT) was building the infrastructure. On July 8, 2025, the AFT, alongside the United Federation of Teachers, lead partner Microsoft, founding partner OpenAI, and Anthropic announced the launch of the National Academy for AI Instruction — a $23 million initiative to provide free AI training and curriculum for all 1.8 million AFT members, starting with K-12 educators[5].
The academy, based at a facility in Manhattan, aims over five years to support 400,000 educators — approximately 10 percent of the U.S. teaching workforce — reaching more than 7.2 million students[5]. It will offer credential pathways and continuing education credits, beginning instruction in fall 2025 and scaling nationally[5].
The partnership is notable not just for its scale but for its structure. It is the first formal partnership between a national teachers’ union and major AI companies, and it places the union — not the tech firms — in operational leadership. AFT President Randi Weingarten framed the arrangement as a question of power: “The direct connection between a teacher and their kids can never be replaced by new technologies, but if we learn how to harness it, set commonsense guardrails and put teachers in the driver’s seat, teaching and learning can be enhanced”[5]. Microsoft President Brad Smith extended the logic: “To best serve students, we must ensure teachers have a strong voice in the development and use of AI. This partnership will not only help teachers learn how to better use AI, it will give them the opportunity to tell tech companies how we can create AI that better serves kids”[5].
OpenAI, serving as founding partner, described the ambition in concrete terms: the academy will “equip 400,000 K-12 educators, about one in every 10 teachers in the US, to use AI and lead the way in shaping how AI is used and taught in classrooms across the country”[6]. The program’s design — workshops, online courses, hands-on training, innovation labs, and feedback cycles — is built around the premise that teachers should help shape AI tools, not merely consume them. As Anthropic Co-founder Jack Clark put it, “how we introduce AI to educators today will shape teaching for generations to come”[5].
The State Legislative Surge: 53 Bills in 2025, 134 in 2026
States have not waited for federal guidance or union academies. During the 2025 legislative session, 53 bills were proposed on AI in education across 21 states, including 51 that applied specifically to K-12 schools[1]. Four states — Illinois, Louisiana, Nevada, and New Mexico — enacted legislation[1]. The pace accelerated dramatically in 2026: according to the tracking firm MultiState, lawmakers filed more than 134 bills across 31 states related to AI in education, focusing on data privacy, classroom use restrictions, literacy, and training[7]. FutureEd, an independent education policy think tank at Georgetown University, is tracking 77 bills across 27 states that specifically address AI in classroom instruction, with 10 already enacted into law as of July 2026[8].
CDT identified five trends across the 2025 bills, with teacher professional development at the top[1]:
- AI literacy and professional development for teachers (15 bills, 1 enacted) — This was the single largest category of state legislation. Tennessee’s SB 677, for example, would have required the state Department of Education to provide a professional development course on classroom AI use and mandated that teachers complete it within two years of certification[1]. New Jersey’s SB 3876 would have required AI literacy integration into K-12 curriculum, age-appropriate sample learning activities, and certificate and degree programs in AI at higher education institutions[1]. The underlying principle: if AI is already in classrooms, teachers need structured, state-backed training — not informal self-education — to use it responsibly.
- Requiring guidance or guidelines on classroom AI use (13 bills, 1 enacted) — New York’s AB 6972 would have created an AI working group to develop model policy and assess the current state of AI in school districts, including understanding tools in use, their cost, and how educators were involved in AI decision-making[1]. Illinois’ HB 2503 would have established an advisory board with standards on safety, privacy, and transparency for AI in education[1].
- Studies and task forces (12 bills, 1 enacted) — New Mexico enacted HB 2, directing the legislative education study committee to create an education data governance and AI working group to study and recommend a structure for improved statewide education data governance and policies to address AI[1]. Massachusetts’ SB 429 would have created a commission to evaluate the current implementation of AI, including assessing the training teachers and students receive on AI and how the school system validates and tests AI systems[1].
- Prohibiting specific AI uses (8 bills, 1 enacted) — Nevada’s enacted AB 406 prohibits public schools from using AI to perform the functions of school counselors, psychologists, or social workers related to student mental health, and requires the state Department of Education to develop guidance for mental health professionals on responsible AI use[9]. Texas’ HB 5282 would have prohibited AI from scoring student assessment responses unless the tool was trained on representative samples, demonstrated validity, and had been independently evaluated[1].
- Addressing AI-generated non-consensual intimate imagery (5 bills, 1 enacted) — CDT research found that 12 percent of students and 8 percent of teachers had heard of deepfake NCII in the 2024-25 school year depicting someone associated with their school[1]. Texas’ HB 1405 would have updated the definition of cyberbullying to include deepfakes, triggering legally required response mechanisms and counseling for victims[1].
The 2026 Enacted Laws: Three Models for Teacher Training
Three of the ten laws enacted in the 2026 legislative session illustrate distinct policy approaches to the teacher training gap:
Idaho S.B. 1227: The Comprehensive Framework
Idaho’s law, signed in March 2026, directs the state education department to develop a comprehensive generative AI framework addressing privacy, procurement safeguards, transparency, academic integrity, AI literacy standards, and professional development[10]. Districts are required to adopt aligned policies. The law also contains an explicit labor protection: it ensures that no AI “replaces or eliminates a human teacher”[10]. Idaho’s approach treats the teacher training question as inseparable from the broader governance question — you cannot train teachers to use AI well without also defining what AI may and may not do in the classroom, how student data will be protected, and where the boundary between human and machine judgment lies.
Maryland S.B. 720: The Infrastructure Model
Maryland’s “Artificial Intelligence Ready Schools Act” takes the most structural approach of any enacted law. It requires the state education department to develop and regularly update AI guidance, requires districts to adopt aligned AI policies and designate AI coordinators, establishes a statewide AI Education Collaborative, supports the evaluation and certification of AI tools, aligns procurement practices with state guidance, incorporates AI literacy into college and career readiness efforts, and provides compensated statewide professional development through a train-the-trainer model[11]. The word “compensated” is significant: Maryland’s law does not merely permit teacher AI training, it funds it and pays teachers for their time. The train-the-trainer model is designed to scale expertise without requiring every district to independently develop curricula. As Maryland State Sen. Katie Fry Hester, the bill’s sponsor, explained: “What we heard repeatedly is that the teachers were feeling like they had to navigate artificial intelligence entirely on their own”[7].
Oklahoma S.B. 1734: The Guardrails-First Model
Oklahoma’s “Responsible Technology in Schools Act” requires every district to adopt a written AI policy before the 2027-28 school year. Policies must address approved and prohibited instructional uses, data protection and minimization, family transparency, and periodic review[12]. The bill requires educator-directed, human-in-the-loop AI use; prohibits AI from serving as the primary basis for grading, discipline, placement, or other high-stakes decisions; and preserves parents’ right to opt students out of student-facing AI tools without academic penalty[12]. Oklahoma’s approach puts the guardrails before the training — it defines what responsible use looks like first, then builds professional development around those norms.
The RAND Finding That Complicates Everything
Beneath the legislative momentum lies a finding from RAND that should give policymakers pause. In interviews with 14 district leaders, RAND researchers found that many had adopted a “do-it-yourself” approach to teacher training after struggling to find external partners to provide appropriate training[2]. District leaders also reported that initial teacher trainings focused on addressing teachers’ fear and discomfort with AI rather than jumping right into instructional tools[2].
This is a critical insight. It suggests that the barrier to teacher AI training is not only funding or legislation — it is the absence of a mature ecosystem of training providers and curricula. The AFT academy, the federal grant priorities, and the state laws are all attempting to build that ecosystem, but they are building it from a starting point where most districts are improvising. RAND also found that low-poverty districts continue to outpace higher-poverty districts in training teachers on AI use[2], raising an equity concern: if AI training scales unevenly, the students who need the most help may end up with the least-prepared teachers.
The Tension Between Speed and Evidence
Justin Reich, an associate professor of digital media at MIT, offered a caution that cuts across all of these efforts. States are trying to regulate classroom AI at a moment when the field is still so unstable that “writing a guide for AI in 2026 is like writing a guide for aviation in 1905” before airlines, airports, or even commercial flight[7]. If you were to evaluate any of the AI literacy or readiness documents against a checklist, Reich said, “there would be a lot of boxes in the ‘we’re making this up’ column and not a lot in the ‘we have evidence’ column”[7].
Yet the alternative to acting without perfect evidence is not inaction — it is allowing the current gap to persist, in which 85 percent of teachers use AI and fewer than half are trained. The RAND report suggests that district leaders themselves acknowledge this tension: many are training teachers not because they have a proven curriculum but because teachers are already using these tools and need immediate guidance[2]. As Noelle Ellerson Ng of the School Superintendents Association noted, AI “was in the classroom the minute students were able to access it”[7]. The policy question is no longer whether AI belongs in the classroom, but whether the adults in the room will be prepared for it.
References
[1] [SECONDARY] Center for Democracy & Technology. “States Focused on Responsible Use of AI in Education during the 2025 Legislative Session.” 2025. https://cdt.org/insights/states-focused-on-responsible-use-of-ai-in-education-during-the-2025-legislative-session/
[2] [SECONDARY] Diliberti, Melissa Kay, Robin J. Lake, and Steven R. Weiner. “More Districts Are Training Teachers on Artificial Intelligence: Findings from the American School District Panel.” RAND Corporation. April 8, 2025. https://www.rand.org/pubs/research_reports/RRA956-31.html
[3] [PRIMARY] Trump, Donald J. “Advancing Artificial Intelligence Education for American Youth.” Executive Order 14277. The White House. April 23, 2025. https://www.whitehouse.gov/presidential-actions/2025/04/advancing-artificial-intelligence-education-for-american-youth/
[4] [PRIMARY] U.S. Department of Education. “U.S. Department of Education Issues Guidance on Artificial Intelligence Use in Schools, Proposes Additional Supplemental Priority.” July 22, 2025. https://www.ed.gov/about/news/press-release/us-department-of-education-issues-guidance-artificial-intelligence-use-schools-proposes-additional-supplemental-priority
[5] [SECONDARY] American Federation of Teachers. “AFT to Launch National Academy for AI Instruction with Microsoft, OpenAI, Anthropic and United Federation of Teachers.” July 8, 2025. https://www.aft.org/press-release/aft-launch-national-academy-ai-instruction-microsoft-openai-anthropic-and-united
[6] [ADVOCACY] OpenAI. “Working with 400,000 teachers to shape the future of AI in schools.” 2025. https://openai.com/global-affairs/aft/
[7] [SECONDARY] Sequeira, Robbie. “As AI use in schools grows, lawmakers and districts scramble to set up guardrails.” Stateline (Pew Charitable Trusts). June 10, 2026. https://stateline.org/2026/06/10/as-ai-use-in-schools-grows-lawmakers-and-districts-scramble-to-set-up-guardrails/
[8] [SECONDARY] DiMarco, Bella. “Legislative Tracker: 2026 State AI in Education Bills.” FutureEd, Georgetown University. Updated July 13, 2026. https://www.future-ed.org/legislative-tracker-2026-state-ai-in-education-bills/
[9] [PRIMARY] Nevada Legislature. “A.B. 406 — Restrictions on Use of Artificial Intelligence for Mental and Behavioral Healthcare.” 83rd Session, 2025. https://www.leg.state.nv.us/Session/83rd2025/Bills/AB/AB406_EN.pdf
[10] [PRIMARY] Idaho Legislature. “S.B. 1227 — Generative AI in K-12 Schools.” 2026 Session. https://legislature.idaho.gov/wp-content/uploads/sessioninfo/2026/legislation/S1227E1.pdf
[11] [PRIMARY] Maryland General Assembly. “S.B. 720 — Artificial Intelligence Ready Schools Act.” 2026 Session. https://mgaleg.maryland.gov/mgawebsite/Legislation/Details/sb0720
[12] [PRIMARY] Oklahoma Legislature. “S.B. 1734 — Responsible Technology in Schools Act.” 2025-26 Session. https://www.oklegislature.gov/cf_pdf/2025-26%20ENR/SB/SB1734%20ENR.PDF
AI automated and generated content. Human reviewed.
Listen to the Podcast

Episode 17 of AI Policy for the Bears. Two bears discuss the AI teacher training gap.
The Weekly AI Policy Brief
AI policy moves fast. Get the week’s developments, new trackers, and podcast episodes in one short email — free, ad-free, no spam.
Leave a Reply