New York City just banned generative AI for students through 8th grade — and whether or not you live in New York, the announcement forced a question into every household: what should MY kids be allowed to do with AI? Here is a practical, age-by-age guide based on the official rules, the built-in parental controls, and what child-development experts actually recommend.
The official age rules (most parents don't know these)
- ChatGPT is not built for children under 13. OpenAI requires parental permission for ages 13–18.
- Most major chatbots (Gemini, Claude, Grok) have similar 13+ policies with parental consent for minors.
- NYC's new moratorium draws its line at 8th grade (roughly age 13–14) — notably consistent with what the platforms themselves say.
So the industry and the biggest US school district roughly agree: under 13, no unsupervised chatbot use.
Set up parental controls (10 minutes)
ChatGPT now ships real parental controls for teens 13–17. From your own account you can:
- Link your account to your teen's and manage it from a parent dashboard
- Set content filter levels — strict, moderate, or standard
- Set blackout hours (e.g., no ChatGPT after 9 pm or during school hours)
- Disable memory and chat history, so conversations don't accumulate into a profile
- Turn off voice mode and image generation if you want text-only use
OpenAI also maintains a parent resource page (chatgpt.com/parent-resources) with age-appropriate defaults. Google, Microsoft, and Anthropic offer family controls with similar scope — check each app your teen actually uses.
Age-by-age recommendations
Under 10 — no direct AI use. This matches the NYC moratorium and platform policies. If your child is curious, use AI together: you type, you read, you discuss. The goal is showing them it is a tool, not a friend.
Ages 10–13 — supervised and specific. Co-use only, for concrete tasks: "let's ask for five science fair ideas." Keep it on your device, in shared spaces. Watch for the two red flags researchers emphasize: treating the chatbot as a confidant, and using it to skip (not check) homework thinking.
Ages 13–15 — managed account, homework rules. Set up the linked teen account with strict filters. Agree on an explicit homework rule — a good one: AI may explain and quiz, never draft. "Explain photosynthesis three ways" is learning; "write my essay" is not, and teachers increasingly detect it anyway.
Ages 16+ — coach, don't police. At this age the goal flips: your teen should be building AI fluency, because the job market they will enter pays a documented premium for it. Talk about what AI is bad at (facts it invents, judgment, originality) as much as what it is good at.
The 5 house rules that work at any age
- AI in shared spaces, not bedrooms — same rule that works for gaming
- The chatbot is a tool, not a friend — say this explicitly and repeat it
- Never share real names, school, address, or photos with any AI
- AI helps you think, it does not think for you — the homework line
- Anything weird, screenshot and show me — no punishment for reporting
The bigger picture
The NYC ban is a pause, not a verdict — the district is spending the year writing age-appropriate rules, and platforms keep adding family controls. The realistic goal for parents is not zero AI; it is supervised, skill-building AI at the right ages. The families that get this right will raise kids who are fluent with the technology and skeptical of it — which is exactly the combination the next decade will pay for.
Sources: OpenAI parental controls announcement and parent resources page; NYC Mayor's Office moratorium announcement (Sep 2, 2026); platform age policies as of September 2026.