New York City just took one of the boldest steps in the U.S. on AI and children. #
Posted September 3, 2026 [ Reviewed by Jessica Schrader
](/us/docs/editorial-process)
Key points
- Keeping AI out of classrooms does not necessarily protect children from the AI they encounter elsewhere.
- Just 90 minutes of AI instruction leaves out critical lessons on judgment, safety, privacy, and relationships.
- Children need to learn how to question AI, verify its answers, protect privacy, and know when not to use it.
On Sept. 2, New York City Mayor Zohran Mamdani and Schools Chancellor Kamar Samuels announced a one-year moratorium on student-facing generative AI for every child from 2-K through eighth grade. That’s nearly 600,000 students, roughly two-thirds of the district. Companion chatbots will be prohibited across all grades, including high school. The city is also placing new limits on screen time, with recommended caps of 30 minutes per day for students in grades 3-5 and 45 minutes for students in grades 6-8.
Developmentally, this makes sense. Children need foundational skills before they can use a powerful tool like AI safely and wisely. They need opportunities to wrestle with a question, sit with not knowing, make mistakes, work through frustration, and build an argument before a machine builds it for them.
Those capacities are developed through effort and struggle. When a tool consistently removes that while these skills are still forming, it's a hindrance.
But the announcement leaves two big questions unanswered:
- What happens when students go home?
- How are we preparing them for a world in which AI will be everywhere?
Removing AI From School Does Not Remove It From a Child’s Life #
According to Common Sense Media’s 2026 census of 1,204 American children ages 9 to 17, 86% already use or interact with AI, and 24% do so every day. Among 9- to 12-year-olds—the very students New York’s moratorium is designed to protect—81% use AI, and 17% use it daily.
Most are not encountering AI at school, either. More than three-quarters use AI on their own phone, computer, or tablet. Only 44% report using it on a school device.
Even if schools decide that students are not ready to use generative AI, that does not mean they are too young to learn about it. It simply means we recognize that AI is an inevitable and ubiquitous part of the world kids live in.
Parents can’t be expected to fill this gap. According to the Common Sense survey, 44% of children say a parent or guardian has never talked with them about how to use AI safely. Even among daily AI users, more than a third have never had that conversation.
So a student can leave an AI-free classroom and walk into a world full of AI—with no teacher nearby and, in many cases, no parent who has ever talked about it. And generative AI is only part of the picture. Artificial intelligence is already embedded in many of the technologies children use every day: YouTube and streaming recommendations, autocorrect and predictive text, navigation apps, search engines, filters, and learning platforms.
Long before children open a chatbot, algorithms are already deciding what they see, what gets recommended to them, and what information rises to the top. We cannot ignore this reality.
AI Literacy Takes More Than 90 Minutes... in High School #
For high school students—the students not covered by the moratorium—New York plans to provide two 45-minute AI literacy modules each year. Two. It actually pains me to type this. The city says those 90 minutes will address what AI is and is not, bias, risks, ethical considerations, and the impact of AI on careers and future skills. These are all important topics, but each requires more than a single lesson. All require foundational knowledge built over time in order for the lesson to stick.
Consider what the city's short list leaves out:
How to verify AI output. Only about half of children surveyed by Common Sense Media say a school or teacher has ever talked with them about how to determine whether information generated by AI is accurate or trustworthy. Verification is not a fact students memorize. It is a habit they develop through repeated practice.Why AI often agrees with them. AI systems are designed to be responsive and helpful. Students need to understand that a confident, agreeable answer is not necessarily a credible one.Who owns what AI creates. Copyright, intellectual property, attribution, and creative ownership have become more complicated in the age of generative AI. Students need traditional media literacy to understand why these things matter.What real human connection requires. New York is right to prohibit companion chatbots for students. But banning a category of technology does not teach children why a machine can simulate empathy without actually caring about them—or what healthy human relationships look like.When not to use AI. This may be one of the most important skills of all. Students should learn to ask:*Will this tool help me think, or will it do my thinking for me?*Knowing when to refrain from using technology is a learned skill, too.How algorithms already shape their lives. AI is increasingly involved in decisions about hiring, admissions, loans, benefits, policing, health, and more. To understand these systems later, students first need to understand concepts such as personal data, biometrics, profiling, recommendation systems, and algorithmic sorting. This is civic literacy.What AI costs. AI has a significant environmental footprint, including energy use, water consumption, and the rapidly growing infrastructure required to support data centers. Many young people care deeply about environmental issues. They deserve to understand this part of the technology too.What to do when something goes wrong. Of children who use chatbots, 17% say one has shown them something they felt was inappropriate or not meant for someone their age, and more than half told no adult. Knowing when and how to seek help is something children need to practice before they are in a difficult situation.How to recognize synthetic media. Deepfakes, voice cloning, nudify apps, AI-enabled sextortion: these are things students are already encountering. They need the skills to recognize manipulation, understand consent, and know what to do when synthetic media is used to harm someone.What they give away when they type in a prompt. Children need to understand that a conversation with an AI system is a conversation with a company collecting personal information about their bodies, health, family, relationships, and emotions.
IntelligenceEssential Reads And this is only a partial list; 90 minutes a year cannot do this work.
AI Literacy Is Not the Same as AI Use #
New York’s moratorium gives educators an opportunity to protect childhood, preserve human-centered learning, and resist the rush to put powerful technology into students’ hands simply because it exists. But keeping AI out of elementary and middle school classrooms cannot mean keeping AI education out too. In fact, this may be exactly the moment to teach it. Students can learn how algorithms work without using a chatbot. They can investigate misinformation without generating it. They can practice protecting personal information without entering it into an AI tool. They can discuss bias, privacy, intellectual property, healthy relationships, environmental impact, and digital ethics without spending a minute looking at a screen. The strongest preparation may be decidedly low-tech: conversation, discussion, role-play, critical thinking, collaboration, media analysis, and opportunities to work through difficult questions with other human beings.
New York’s 600,000 children will spend this school year living in a world shaped by AI. Many will use it on their own devices, in their own bedrooms, without a teacher nearby and, for nearly half of them, without a parent who has ever talked with them about how to use it safely. The question is not simply whether children should use AI in school. It is whether we are preparing them with the knowledge, judgment, and human skills they will need to live with it everywhere else.
References
O’Neil, E., Anderson, K. S., White, B., Zimmermann, L., Mann, S., & Robb, M. B. (2026). The Common Sense Media Census: AI Use by Tweens and Teens, 2026. Common Sense Media.