New York bans generative AI in elementary schools for a year, and the debate already divides those who ran its own schools

By Zendoric · September 7, 2026.

New York has decided to ban the use of artificial intelligence in public school classrooms up to high school for at least this school year, which begins on September 10. The measure, announced this week by Mayor Zohran Mamdani, softens an earlier proposal from the City Council, which had called for a two-year moratorium on generative AI, citing insufficient data privacy protections and a lack of research in the Department of Education's draft guidelines, as reported by CBS News New York.

The debate went public thanks to CBS New York's program "The Point with Marcia Kramer", where former schools chancellor David Banks (who led the system from January 2022 to mid-October 2024) and Richard Buery Jr., current chief executive of Robin Hood — the city's anti-poverty charity — laid out opposing positions. Banks openly disagrees with the moratorium: "AI is here, it's everywhere, our children interact with it whether or not they do so in school", he argued. His objection is not to guardrails themselves — he says he understands them — but to treating them as a uniform pause: he calls for distinguishing between a second grader and a tenth grader, and warns that "two years from now we'll be even further behind". Buery, for his part, defended a three-point plan to educate families about the technology, without the article detailing its content.

The actual figure is worth clarifying: what takes effect is not the two years the Council requested, but "at least" one year, according to CBS itself. It is a meaningful difference, because it marks a short, reviewable extension rather than a long moratorium settled on shaky ground.

Our reading is that this New York episode — the largest public school district in the United States — precisely illustrates the tension that defines AI deployment in public institutions: regulatory caution is legitimate when guarantees on minors' data privacy and solid pedagogical evidence are lacking, but banning it in the classroom does not ban it at home. Households with resources will keep giving their children access to AI tutors outside school hours; those without them will not. A moratorium designed to protect the most vulnerable students may end up widening the very gap it set out to close, if it is not accompanied by a subsequent adoption plan with a clear timetable.

This connects with a thesis we have been arguing about AI and education: the value lies not in banning or adopting wholesale, but in what role the teacher plays. The teacher who learns to orchestrate the tool — turning it into an augmented tutor, adapted to each age and level — wins, and the model that merely transmits content loses, with or without AI in the mix. The distinction Banks calls for between a second grader and a tenth grader points exactly there: the problem is not technology in the abstract, it is the absence of a graduated pedagogical framework that most districts, not just New York, still lack.

As sector context, most education authorities in the United States and other countries are improvising generative AI policies with barely any longitudinal data on their effect on children's learning, which makes it reasonable to demand privacy safeguards before use becomes widespread. But Banks's warning has substance: the gap that opens during the pause is not neutral, and "waiting for certainty" is not a cost-free position. In the long run, AI's real promise in education — personalized, affordable tutoring for any student, not just for those who can pay for it — is exactly what a poorly calibrated ban may delay more in public classrooms than in private ones. Short-term caution is justified; the challenge is that it does not inadvertently become another form of inequality.