WHAT THE POLICY SAYS
● Pre-K to 8th grade: no student-facing generative AI at all. Chatbots and AI tutors both.
● Pre-K to 2nd: no individual devices.
● Screen time caps: 30 minutes in grades 3 to 5, 45 minutes in 6 to 8.
● High school: AI reserved for literacy, career readiness and five pilot schools.
The scale of it
New York City's Department of Education runs the largest public school system in the United States. A policy this restrictive, from this district, sets a reference point that other districts will cite whether they follow it or not.
And the restriction is more specific than a general ban. It targets student-facing generative AI, which leaves teacher-facing use open — lesson planning, differentiation, administrative work. What is barred is the child talking to a model.
The AI tutor exclusion is the notable part
TUTORS WERE MEANT TO BE THE GOOD CASE
Personalised tutoring has been the strongest argument for AI in education — patient, available, adaptive to a child's pace. It is what most edtech funding has chased.
Naming AI tutors explicitly alongside chatbots means the district considered that argument and rejected it for younger children, rather than banning a category it had not thought about.
The reasoning is not published in the summaries available, so this is inference: the screen-time caps sitting alongside the AI restriction suggest the concern is about developmental appropriateness and device dependence, not only about accuracy or cheating.
What it does allow
| Group |
What is permitted |
| Pre-K to 2nd grade |
No individual devices, no student-facing AI |
| Grades 3 to 5 |
Devices capped at 30 minutes, no student-facing AI |
| Grades 6 to 8 |
Devices capped at 45 minutes, no student-facing AI |
| High school |
AI for literacy and career readiness, plus five pilot schools |
The five pilot schools are the interesting clause. A district this size running a controlled pilot rather than a blanket rollout produces exactly the evidence the debate currently lacks — and the pilots are where anyone building education AI should be paying attention.
The argument on both sides
For the restriction: children in primary school are learning to read, reason and sit with difficulty. A tool that supplies answers fluently interferes with the part where the learning happens. There is also no strong published evidence that AI tutors improve outcomes for that age group, and the screen-time literature is not encouraging.
Against it: the students who benefit most from a patient, always-available tutor are the ones whose families cannot buy human tutoring. A ban that applies to everyone equally does not affect everyone equally. And children will use these tools outside school regardless, so a school that never teaches AI literacy has not prevented use, only supervision.
Both arguments are serious, and the pilot structure is a reasonable way to find out which is right rather than assuming.
What it means practically
| If you are... |
The read |
| Building education AI |
The addressable market for under-14 student-facing tools just narrowed in the largest US district. Teacher-facing did not |
| A parent |
This governs school use. Home use is your decision, and the policy is a reasonable prompt to think about it |
| A teacher |
Teacher-facing use is not what is restricted. Planning and administrative tools are unaffected |
| Watching AI policy generally |
This is the first large-district policy specific enough to copy. Expect it cited in other districts within months |
Sources
FAQ
What did New York City ban?
Student-facing generative AI, including chatbots and AI tutors, in pre-K through 8th grade, under a policy released on 3 September 2026.
Does it apply to teachers?
The restriction is on student-facing use. Teacher-facing applications such as planning and administration are not what is barred.
What about high school?
AI access is reserved for literacy, career readiness and five pilot schools rather than being generally available.
What are the screen time limits?
Individual devices are capped at 30 minutes in grades 3 to 5 and 45 minutes in grades 6 to 8. Pre-K through 2nd grade have no individual devices at all.
Why ban AI tutors specifically?
The reasoning is not detailed in the available summaries. Naming tutors alongside chatbots indicates a considered decision rather than an oversight, and the accompanying screen-time caps suggest developmental concerns rather than only accuracy ones.
Does this affect other districts?
Not directly, but New York City runs the largest public school system in the United States, and a policy this specific tends to be cited as a reference point elsewhere.