Every Edition of the AI in Public Education Brief

California AB 1159 became law on September 10, 2026, prohibiting operators from using student data to train or develop AI systems and giving a harmed pupil the right to sue. One day earlier the AFT, the UFT, and Microsoft announced a National AI Safety and Privacy Standard, a contractual benchmark districts can request that covers only products built for education, with Microsoft promising nationwide reach from November 1. SB 1128 bars mandatory take-home devices in kindergarten from 2027-28. A peer-reviewed study found twelve hours of AI literacy instruction did not significantly move a validated knowledge measure for 36 high school students. Brief Edition 33.

New York City Public Schools announced a one-year moratorium on student-facing generative AI for grades 2-K through 8 on September 2, 2026, affecting nearly 600,000 students, with companion chatbots prohibited at every grade and five named pilots capped at 50,000 high school students. The same week, Los Angeles Unified confirmed that its web-filtering software blocks AI tools for all students at all grade levels on district-issued devices, roughly 378,000 students, without a board vote. An IBM survey conducted by Morning Consult found 76 percent of middle school educators using classroom AI at least weekly against 20 percent reporting extensive training. Brief Edition 32.

The Utah State Board of Education, with Brigham Young University and Internet Safety Labs, captured live network traffic from 100 education apps and found that in 44 of the 85 apps carrying signed data privacy agreements, the app collected at least one data element the agreement did not permit. California AB 2392, requiring generative AI procurement standards and training for its public colleges, cleared the Legislature on August 26. The Fulton County Board of Education in Georgia carried Board Policy IFBI limiting AI output to suggestions and requiring human final decisions. Brief Edition 31.

On August 20, 2026, the Office of Elementary and Secondary Education issued a Dear Colleague Letter telling states and districts to evaluate education technology on demonstrated learning outcomes rather than screen time alone, to build evidence of effectiveness into procurement and renewal decisions, and to look for products that have conducted randomized controlled trials. Ten days earlier, the Institute of Education Sciences reported that a search of the What Works Clearinghouse identified no research studies on AI and student outcomes. California passed SB 1159, writing into the Brown Act that an AI system is not a member of the public, and sent the Governor SB 928, requiring that an instructor of record be a person. Brief Edition 30.

On August 11, 2026, the Albuquerque Journal reported that the New Mexico Public Education Department had issued updated guidance offering districts three ways out of Amira, the AI reading assessment the state requires, after Santa Fe, Los Alamos, Farmington, Roswell, Clayton, and Turquoise Trail Charter School declined to run it over its voice recording of children. The exemptions last one school year. The same week, a federal court left standing the premise that the district, not the parent, is the consenting party for classroom software, and new peer-reviewed research found chatbot-assisted inquiry lifted problem-solving but not critical thinking, because it removed the verification step.

On August 3, 2026, AASA announced that students representing all 50 states, convened as a National AI Student Senate at the Edward M. Kennedy Institute, had drafted, debated, and passed the STUDENTS FIRST Act of 2026 by a vote of 82 to 16: due-process rights on AI-misuse accusations, human review of consequential decisions, profiling prohibitions, data-collection disclosure, and the right to decline AI with an alternative assignment. AASA is circulating the framework to more than 10,000 school leaders. The same week, two German longitudinal studies found adaptive tutoring licenses produced no class-level learning gains, and a six-district Colorado study found AI policy built by improvisation.

On July 21, 2026, the Florida Department of Education noticed a draft amendment to Rule 6A-1.0957 that would fold AI governance into the internet safety policy every district and charter board already re-adopts annually: opt-in parental consent with a non-AI alternative, a public list of approved tools, teacher training, interaction records, and usage reporting to the state, all by January 1, 2027. The ed-tech trade group SIIA filed objections July 29. The same week, a nationally representative study found nearly two-thirds of student handbooks never mention AI.

On July 18, 2026, the American Federation of Teachers adopted a sweeping artificial intelligence resolution at its convention: no screens, including online assessments, in prekindergarten through second grade, opposition to student-facing AI in elementary schools, a ban on social companion chatbots at minimum until age 16, and an enforceable right to bargain over AI procurement, implementation, and oversight. The same window, a study of 424 districts across three states found most district AI policies provide minimal direction, with policy quality tracking district wealth.

On July 10, 2026, Illinois enacted Public Act 104-0565, barring an evaluator from using artificial intelligence to assign a numerical score or qualitative rating for any component of a teacher's evaluation, while expressly permitting AI to support administrative tasks. It passed the Senate 55-0 and the House 113-0, and takes effect January 1, 2027. The same week, Hawaii enacted disclosure and crisis-protocol requirements for conversational AI used by minors, and a peer-reviewed experiment found that the design of an AI report shifts teacher judgment even when the student work is unchanged.

On Monday, July 6, 2026, New York City Schools Chancellor Kamar Samuels asked principals to hold off on educational software purchases until the Education Department finalizes its AI guidance. At a June 24 City Council hearing, officials could not say how many schools were using AI products or which ones, because purchasing is not tracked centrally. The same day, Illinois enacted Senate Bill 315, the first state law requiring independent third-party safety audits of the largest AI systems. New research: institutional AI readiness reaches students only through teacher capability, not enthusiasm, and four widely used language models changed their writing feedback for identical eighth-grade essays based on the student's stated race, learning needs, or achievement level.

On June 30, Governor Newsom signed Assembly Bill 2148, amending the California Education Code to define a public school employee and a contractor providing services in a public school as natural persons, the first clear instance of a state writing a human floor into K-12 education code. Parallel measures reach higher education in California and Illinois. The strongest K-12 evidence this cycle points the same direction: a preregistered 2,003-student randomized controlled trial of an adaptive math tutor found no overall effect, a 28-study systematic review finds positive but mitigated tutoring effects, and AI-dependence research links heavier reliance to weaker critical thinking, only partly offset by literacy.

On June 25, California released the model AI policy required by Senate Bill 1288, handing the nation's largest school system a state template; three days later, on July 1, Ohio's statutory deadline for every district, community school, and STEM school to adopt an AI policy takes effect. A model policy is a floor and a deadline is a due date, and neither by itself is governance. New research: commercial AI models do not mirror teachers' own views of AI, the newest classroom-design paper proposes a multi-year trial that has not been run, and a peer-reviewed Delphi panel agrees on only 8 of 23 statements defining AI literacy.

The U.S. Senate held its first K-12 AI hearing on June 16, and the witnesses asked Congress to fund teacher training and research and protect E-Rate, not to set a national standard. The K-12 AI Literacy and Readiness Act (H.R. 8747) and the LIFT AI Act (S. 4414) free existing federal dollars for AI without setting a floor. New research: AI-tutor access produced almost no use and no reading gain, teacher professional development is the lever that moves instructional quality, and a witness tells the Senate the long-term causal evidence does not exist.

Connecticut enacts Public Act 26-15, folding AI-inclusive computer science into required instruction and pairing it with minor-safety rules, while Florida's AI Bill of Rights dies in the House and New York moves to keep most AI out of K-8 classrooms. A peer-reviewed meta-analysis finds AI agents lift skills and knowledge but not, at significance, higher-order thinking; two studies find off-the-shelf LLMs are not grade-appropriate for younger learners.

Maryland enacts the Artificial Intelligence Ready Schools Act (SB 720, Chapter 634), making a non-instructional district AI coordinator a statutory mandate in every local school system and standing up a state AI Education Collaborative. New Mexico weighs a formal oversight body after its own analyst calls 2025 guidance non-binding. Peer-reviewed research: awareness and ethics, not technical skill, predict academic performance.

Pennsylvania AG files first state executive action against a consumer AI chatbot for impersonating licensed psychiatrists. Federal K-12 AI Literacy and Readiness Act would re-permission Title I, II, IV funds. Houston ISD quadruples Future 2 to nine campuses. Peer-reviewed PNAS: unrestricted GPT-4 caused 17 percent grade decline when removed.

Colorado SB 189 repeals its AI Act with a FERPA carve-out. The Lawrence Public Schools federal surveillance case advances on Fourth Amendment grounds. CRPE Early Adopters database shows tracked AI-adopting districts doubled year over year.

Michigan published a six-practice framework. Tennessee restricted AI from mental health roles with a private right of action. Two state actions, one convergence point for districts.

The largest U.S. school district closed AI guidance comment, three peer-reviewed studies redefined what AI tools must demonstrate, and the federal AI priority takes effect May 13.

FTC COPPA enforcement, a Florida criminal probe of OpenAI, and seven Tumbler Ridge lawsuits redefine AI vendor accountability for school districts.

Maryland’s AI Ready Schools Act, Stanford’s evidence gap, and the OECD performance-learning paradox converge on one governance challenge district leaders cannot defer.