New York halts generative AI through eighth grade: a moratorium and hard caps on screen time

On 2 September 2026, the Office of the Mayor of New York announced a moratorium on student-facing generative AI and on companion chatbots in the city's public schools. The measure, described in the official press release as the broadest of its kind in the United States, takes effect with the 2026-27 school year and covers the grades from pre-K through eighth. It reaches nearly 600,000 students, roughly two thirds of those enrolled in the country's largest school system. In the official release of 2 September 2026, Mayor Zohran Mamdani says that “children need teachers and human connection to learn and grow.” The plan also sets tight caps on individual screen time: no individual use through second grade, a maximum of 30 minutes a day in grades three to five, and up to 45 minutes in grades six to eight.

For high schools, the official release allows limited, guided use inside pilot programs that will involve at most 50,000 students (about 5% of the public school population), plus two 45-minute modules a year on critical thinking about AI for every high school student. Exemptions are carved out for assistive technology for students with disabilities, for tools aimed at multilingual learners, and for career preparation pathways such as computer science. According to the independent education outlet Chalkbeat New York, routine computer-based reading and math assessments also fall outside the moratorium. For teachers, AI remains allowed for work such as lesson planning, translation and drafting communications, while it is expressly barred from student evaluation, behavioural monitoring, counselling and the writing of individualized education programs. Chalkbeat also reports that the decision follows pressure from parent groups and a letter signed by more than half the members of the City Council, asking the mayor and the chancellor to suspend the use of AI in schools.

On how the policy will actually be enforced, though, the available documents say little. Chalkbeat notes that the ways in which the education department will hold schools to the guidance remain undefined. The two sources also describe the high school piece in terms that do not line up: the Mayor's release gives a ceiling of 50,000 students in the pilots, Chalkbeat speaks of five pilot programs, and the available documents make it impossible to establish whether the scope is the same. There is no documentation of which concrete tools fall under the definition of “student-facing generative AI” — whether, for instance, it covers AI features already built into teaching platforms in use — and no criteria have been published for deciding whether the measure will be renewed beyond the current school year. One last note: the official guidance page on the New York City public schools portal did not return the text of the document during our checks, so the operational details reported here rest entirely on the Mayor's office release and on Chalkbeat's reporting.

Until now, rules on generative AI in schools were almost everywhere responsible-use guidelines rather than bans. New York chooses instead to reverse the burden of proof, treating AI not as an inevitable evolution but as a variable to be isolated and measured. In the official release of 2 September 2026, Chancellor Samuels stresses that “innovation does not mean more technology, and over the coming year evidence will be our guide.” The open question is the one Chalkbeat flags and neither document settles: who checks that the moratorium is respected, and with what tools.

— Olya