The largest school district in the United States has decided its youngest students can wait a year before meeting a chatbot. New York City announced a one-year moratorium on student-facing generative AI for children in 2-K through eighth grade, effective in the 2026–27 school year. The pause covers nearly 600,000 pupils — roughly two-thirds of the system’s enrollment — and was unveiled on September 2, 2026 by Mayor Zohran Mamdani and Schools Chancellor Kamar Samuels.
The policy removes all software with student-facing generative AI from those grades, and goes one step further with companion chatbots, which are prohibited across all grades. “The tech industry wants us to believe that A.I.-powered early education is not only inevitable, but necessary. We do not see it that way,” Mamdani said.
High schools are largely exempt, but under tight conditions:
- Five AI pilot programs, capped at five classes per school, open to up to 50,000 students — around 5 percent of the public school body
- Weekly usage limits that vary by tool
- Twice-yearly AI literacy modules of 45 minutes each for high schoolers
New York is not improvising here. Norway banned generative AI in its primary schools — grades 1 through 7 — in June 2026, with the ban taking effect at the end of August. Prime Minister Jonas Gahr Støre argued that AI raises the risk of children skipping essential steps in learning to read, write and do mathematics.
Europe’s broader rulebook, meanwhile, comes at the problem from the opposite direction. The EU AI Act classifies educational uses of AI as high-risk under Annex III — but the uses it targets are admission decisions, evaluation of learning outcomes, assessment of educational level and exam monitoring. In other words, it regulates AI applied to pupils by institutions, not AI used by them. Compliance with the Annex III obligations is due by December 2, 2027, while the Act’s Article 4 AI literacy requirements have already been binding since February 2025. Italy has gone furthest on the age question: its Law 132/2025, in force since October 2025, requires parental consent before children under 14 can use AI.
The pattern is hard to miss. Within a few months, two governments on opposite sides of the Atlantic concluded that generative AI and early education should be kept apart until someone can show the technology helps children more than it hurts them. New York’s moratorium is explicitly framed as a year to study the impact on learning and development — which means the 2027 decision, whichever way it goes, will be one to watch far beyond the five boroughs.
