NYC Banned AI in Schools. That Won't Work.

This column was originally published in EdTech Digest.

On September 2, Mayor Mamdani announced a one-year moratorium on student-facing generative AI in New York City public schools, covering pre-K through eighth grade. About 600,000 students are affected. It is the broadest restriction of its kind in the country.

I understand the impulse. Teachers are overwhelmed. Parents are worried. Nobody has a consensus on what responsible AI use looks like in a fourth-grade classroom. A moratorium buys breathing room.

But breathing room is all it buys.

A year from now, the moratorium will expire, and the same questions will be waiting: Which tools are appropriate? How should students interact with them? What does cheating look like when a chatbot can write a five-paragraph essay? None of those questions answer themselves while AI sits behind a firewall.

Meanwhile, the students this policy is meant to protect will keep using AI at home. The ones with tech-literate parents will learn to use it well. The ones without will fall further behind. Private schools in the same city face no such restriction. The moratorium doesn't eliminate AI from students' lives. It just makes the gap between who learns to use it and who doesn't a function of family income.

The real problem is upstream

Most teachers have never been trained on how large language models work, what they are good at, where they fail, or how to set boundaries around them in a classroom. That isn't a criticism of teachers. Until recently, there was no reason for an eighth-grade English teacher to understand token prediction or hallucination rates. Now there is, and most professional development hasn't caught up.

Bans are a response to that gap. Literacy is the fix for it.

When a teacher understands how AI generates text, she can design an assignment that uses it as a drafting partner without replacing the student's thinking. When she understands its tendency to fabricate sources, she can teach students to verify claims instead of trusting output blindly. When she knows the difference between a school-safe platform with audit trails and a consumer chatbot with no guardrails, she can make an informed decision about what belongs in her classroom and what doesn't.

That teacher doesn't need a moratorium. She needs training.

What happens when teachers get it

We've seen what happens when teachers get it. After 8,500 educators completed an AI literacy certification we built, the pattern was consistent: the teachers who were most skeptical going in became the most intentional users. They didn't adopt AI wholesale. They adopted it on their terms, with clear boundaries, for specific purposes. One AP U.S. History teacher built AI tutors that coach her students through thesis statements and DBQ essays, giving each kid individualized feedback she couldn't scale on her own. Her students didn't just tolerate the change. They started requesting the AI-integrated lessons.

That isn't a story about technology. It's a story about a teacher who understood the tool well enough to wield it deliberately.

Teacher literacy scales

The good news is that teacher literacy scales. A two-hour, self-paced course can cover the fundamentals: how models work, where bias enters, what student data privacy requires, how to evaluate AI output, and how to write a classroom AI policy. It's not a graduate degree. It's the minimum viable knowledge a teacher needs to make a confident, informed decision about AI in her classroom.

That is exactly what districts should be investing their moratorium year in. If New York spends the next twelve months training every teacher on AI fundamentals, the moratorium will have been a bridge to something durable. If it spends the year keeping AI out and hoping the questions resolve themselves, it will end up in the same place — a year later, with the same gap.

The pattern here isn't new

The pattern here isn't new. When the internet arrived in schools, the first response was restriction. Schools blocked websites, banned search engines, worried about what students would find. The districts that invested in digital literacy early built the most capable students and the most confident teachers. The ones that only blocked fell behind.

AI is following the same arc, faster. The schools that train their teachers now will be ready when the restrictions lift. The ones that don't will still be debating whether to allow AI while their students are already using it to study for the SATs.

Bans expire. Literacy doesn't.