Schools are banning AI for kids faster than anyone is regulating it
NYC and LA just restricted student AI use, but no privacy law yet covers what these classroom tools actually do with children's data.
What happened: New York City and Los Angeles, the two largest US school systems, adopted strict new AI rules for the 2026-27 school year. NYC's one-year moratorium blocks generative AI for students through eighth grade, allowing only limited, supervised use in some high school classes. Los Angeles Unified temporarily blocked all grades from accessing generative AI tools. Other districts went the opposite way: Dallas ISD is training teachers and promoting responsible student use, while Houston ISD launched an AI-powered learning pilot at two elementary schools. The split follows a year in which several states curbed phone and social-media use in classrooms over similar parent and teacher concerns.
Why it matters: No single law governs how schools can use AI with children. Researchers at the University of West Florida argue FERPA, the main federal student-privacy law, was never built for tools that track behavior and generate predictive judgments about kids, so AI-based monitoring can spread with little oversight. A separate research team found unclear lines between federal and state responsibility, murky rules on commercial use of student data, and little transparency into how school algorithms make decisions. Critics also warn outright bans could backfire: students without AI access at home may fall further behind peers who get guided, supervised practice elsewhere.
What the research says: Reviews broadly agree technology alone doesn't improve learning. A Brookings Institution analysis found adding AI hasn't moved outcomes by itself, and concluded that right now, risks to children's cognitive and social development outweigh proven benefits, though the authors remain optimistic AI could eventually help if it supports rather than replaces teaching. Two Czech researchers catalogued real upsides like personalized pacing and lower academic stress, weighed against loneliness, "technostress," and weaker face-to-face skills. One Kansas scholar argues the deeper issue is that schools still reward memorization-style tasks AI can already do, so bans treat a symptom rather than fixing how schools assess learning.
The policy response: Fixes are emerging piecemeal rather than as one framework. Massachusetts issued state guidance telling districts to align AI use with privacy, accessibility, data-governance, and accountability rules. In Washington, a 2025 executive order created a White House task force on youth AI literacy, and two bipartisan bills, one pushing state AI-in-education standards, another adding privacy protections and parental controls, remain stalled in Congress. Researchers are pushing to update FERPA itself to cover predictive profiling and require independent audits of software districts buy from private vendors.