NYC Banned AI From 600,000 Classrooms. Here Is the One Skill Worth Building Before the Rest of the Country Follows.

New York City just banned AI from classrooms for about 600,000 students in pre-K through eighth grade. The city called it the nation’s most comprehensive AI moratorium in K-12 education. I read through the announcement and the reasoning behind it, and I kept thinking about the same thing: the school ban only applies at school. At home, on devices, nothing changes. The question the ban raised for me, as a dad, was not whether it was the right policy call. It was what parents are supposed to do in the part of the day schools cannot reach.^[1]

New York City, the nation’s largest school district, banned student-facing AI for about 600,000 students in pre-K through eighth grade, calling it the broadest AI moratorium in K-12 education in the country.

On September 2, 2026, Mayor Zohran Mamdani and Schools Chancellor Kamar Samuels announced the policy, which takes effect for the 2026-27 school year. Students in pre-K through second grade will have no individual devices at all. Grades three through five get a 30-minute daily screen cap. Grades six through eight get 45 minutes. Across all those grades, student-facing AI tools are fully banned.^[1] The policy affects about two-thirds of New York City’s public school enrollment.^[2]

The reasoning from city officials was specific. “Children need teachers and human connection in order to learn and grow,” Mayor Mamdani said when announcing the policy.^[2] Chancellor Samuels added that “innovation does not mean more technology,” and committed to spending the year gathering evidence before deciding how to extend any AI access to younger students. The city’s position was that younger students need what screens and AI tools cannot provide: hands-on learning, real social interaction, and instruction led by a person in the room.^[1]

New York City is not the only district rethinking this. Cities and states across the country are implementing new AI and screen time restrictions as the 2026-27 school year begins, citing concerns about student learning, screen exposure, and what young learners lose when AI handles more of their schoolwork. The pattern is national. Something is shifting in how school systems are weighing the tradeoffs of early AI access for children.

The premise of the NYC ban is that young learners need to build cognitive habits through struggle, repetition, and real effort before those habits can be extended with AI. The school ban protects that process in classrooms. It cannot protect it at home.

Children in elementary and middle school are in the process of forming the habits they will carry for life: how to organize their thoughts, how to push through confusion, how to revise something and notice why it needed changing. These habits build through repetition, and they require doing the work yourself. When AI does the hard part, the repetition still happens, but the skill does not develop. The child produces the output but skips the process that builds the capacity.

Research has begun to show the effects on the youngest learners. A Canadian longitudinal study of nearly 2,500 toddlers found that higher screen exposure was linked to missed developmental milestones on developmental screening tests, particularly for children in the earliest years.^[3] NYC’s decision to remove individual devices entirely for the youngest grades reflects this finding directly. The city is not just restricting AI. It is protecting the conditions under which foundational learning happens.

The school ban will not change what happens at the kitchen table. Your child can still open a chatbot before they open a blank document. They can paste a homework prompt into AI before writing a single word. Nothing in the NYC announcement reaches those moments. The decision about what happens at home belongs entirely to parents. And most parents have not had a framework for thinking about it until now.

The skill is Earned Access: before a child uses AI for a task, they demonstrate they can do the task without it, even imperfectly, so the AI becomes a collaborator rather than a replacement.

The principle is not new. It was built into how school worked before AI existed. You were not allowed to use a calculator in arithmetic class until you had shown you understood the math. You were not allowed to look up answers during a spelling test until you had practiced enough to make the answer meaningful when you found it. The tool was available, but access to it was sequenced behind the skill itself. That order was the point.

Earned Access is the same principle applied to AI. Before a child uses an AI tool to help write an essay, they write a draft. It does not have to be good. It just has to be theirs. Once they have made the attempt, they can use AI to compare, challenge, or extend what they wrote. The AI becomes a collaborator on something they already own, not a substitute for the thinking they skipped. The sequence matters more than the tool.

What happens when the sequence gets reversed is subtle and easy to miss. If a child offloads a skill to AI before internalizing it, they do not lose time. They lose the capacity itself. The brain builds the habit of reaching for AI before building the habit of doing the task. I think about my own daughters. My 13-year-old is fast and disciplined. When she gets an assignment she does not want to think through, she reaches for a shortcut quickly. My 9-year-old will still sit with a problem for a few minutes. My 7-year-old gets stubborn about figuring things out alone. That stubbornness does not bother me. I want to protect the instinct underneath it.

A child practicing Earned Access tries to work through problems independently before reaching for help, and can describe how they got to an answer, not just show you the result.

The clearest sign is what they do in the first few minutes of a hard task. They stay with it. They might complain. They might stare at a blank page for a minute before writing something imperfect. But they stay. My youngest sat with a math problem last week for four minutes before she figured it out. I watched and said nothing. She was genuinely proud in a way she would not have been if I had stepped in at the thirty-second mark. The staying is the practice. Four minutes of staying is four minutes of building.

They can also walk you through what they tried. Not just the finished answer. The path. “I tried this first and it did not work, so I tried something else.” That narration means the child owns the process, not just the output. A child who can only show you what the AI produced has not learned anything from completing that task. A child who can show you their attempt, even a failed one, and explain what they were thinking, has done the work that matters.

Pick one task where your child currently reaches for AI first, and this week require them to make a genuine attempt alone before any AI use is allowed.

Choose one assignment or subject where your child has been defaulting to AI. Tonight, set the rule: the first attempt must come from them. It does not have to be finished or correct. It just has to represent what they can do on their own. After they have made the attempt, they can use AI to compare their answer, see what they missed, or take their draft further.

Ask your child what AI tools they have access to through their school software. Many families do not know that AI is already embedded in tools their kids use for school every day. Knowing what your child has access to during school hours helps you understand where the Earned Access habit is most worth building at home, and what “doing it yourself first” means in practice for each subject.

Frame the expectation as a privilege, not a punishment. Tell your kid: “I want you to know what you can do first, so the AI actually makes you better instead of replacing you.” A child who understands the reason behind the rule is more likely to apply it on their own, even when no one is watching. That internalization is the whole goal.

New York City will spend the year gathering evidence on what the moratorium actually shows. I do not know what they will find. What I do know is that the instinct behind the decision is right. Something gets built in a child who has to work through hard things without a shortcut. It is hard to name precisely and easy to lose before you notice it is gone.

You do not need a ban at home to protect it. You just need to change the order. Show me what you can do first. Then let the tool help you do it better. That is the whole idea behind Earned Access, and that is what 600,000 families in New York City are now depending on their schools to enforce. You can start it at home today, before anyone makes a policy about it.