In early September, New York City announced a one-year moratorium on student-facing generative AI for children from pre-kindergarten through eighth grade, affecting close to 600,000 students in the largest school system in the country. The policy has been widely described as the broadest of its kind, and it arrived just as many districts were finalizing their own AI guidance for the school year. School leaders elsewhere are now asking a practical question about whether this signals that AI has no place in a teacher's workflow. The answer depends on distinguishing between tools students talk to and tools teachers use.

The moratorium targets software that lets students generate text, images, or other content in response to their own prompts, along with companion chatbots at every grade level. Officials said the aim is to protect reading, writing, and reasoning for younger children while the city studies the effects of the technology over the 2026-27 school year. A new coalition of educators, parents, union partners, and advocates will review the results and publish recommendations. High schools may run limited pilots and will offer AI literacy lessons, so the policy is better understood as a cautious pause than a blanket rejection.

Teacher-side use is treated differently in the policy, though the details deserve a careful read. Teachers may continue to use approved tools for planning and operational tasks that meet the city's safety standards, while press reports indicate that asking an AI system to grade assignments is not part of what is permitted. Any district considering AI-assisted grading should therefore read its own governing policy closely rather than assuming that permission for planning extends to scoring student work. That distinction between preparing lessons and evaluating students is likely to appear in many other district policies this year.

Student-facing versus teacher-facing AI

The core distinction is who interacts with the model and what comes out of it. A student-facing tool puts generated language in front of a child who is still learning to write, which raises concerns about dependence and about whose thinking appears on the page. A teacher-facing tool that drafts rubric-aligned comments for review never shows output to a student until a teacher has read, edited, and approved it. Policymakers may eventually draw very different lines around those two uses, and school leaders should plan for that possibility.

• Read your district policy for the exact wording on grading, scoring, and feedback.
• Confirm whether the policy covers tools used only by staff or only those used by students.
• Ask vendors to explain what student data is shared and whether students ever see raw model output.
• Keep a written record of where a teacher reviews and approves each piece of AI-drafted feedback.
• Plan to revisit decisions once the city's coalition publishes its recommendations.

Grades 6 through 8 and the writing classroom

Middle school ELA teachers in affected schools face a particular tension. They still need to assign and grade substantial writing, and the policy elevates writing and reasoning as priorities, yet their grading load does not shrink because students are not using chatbots. A seventh grade teacher with 150 students who assigns a two-page literary response every other week faces well over a thousand essays in a single semester. Without any technology help, that volume pushes feedback turnaround toward two or three weeks, which is too slow for students to connect comments to their writing.

Where local rules allow teacher-side tools, the sensible approach is to keep the model behind the teacher. That means rubrics written by the teacher, scores treated as suggestions, and comments reviewed before release. Where local rules do not allow it, teachers can still gain time through comment banks, shared rubrics, and focused feedback on one or two criteria per assignment. Either way, the policy debate is a reason to document workflows more carefully, not a reason to stop improving them.

Questions school leaders should ask this fall

Leaders in other districts do not need to copy New York's decisions, but they do need to answer the same questions on purpose. Which grade bands will allow student-facing AI, if any, and under what supervision? Which teacher uses are approved, and who reviews tools before staff adopt them? Writing these answers down before an incident occurs prevents a patchwork of individual rules from forming classroom by classroom.

Recent survey work on AI in schools found that fewer than half of principals reported having any AI policy, and many teachers had never seen the guidance that did exist. That gap means a clear one-page staff policy can be a meaningful improvement on its own. The page should name approved tools, define what human review means, and say who to contact with questions. A document that short is more likely to be read and followed than a forty-page framework.

Keeping teacher judgment at the center

Whatever the specifics, the safest and most defensible position is that AI never replaces teacher judgment in assigning grades. A rubric-based workflow keeps judgment where it belongs: the teacher defines the criteria, the system proposes a first-pass score and draft comments against those criteria, and the teacher decides what students receive. That structure is easy to explain to parents, administrators, and unions, because accountability stays with a named professional. It also leaves room to adapt if a district narrows what is allowed.

New York's pause will likely shape the conversation for months, and other districts will cite it in both directions. Schools that have already defined their human-in-the-loop expectations will be better prepared to respond, whether the eventual guidance tightens or loosens. The most useful preparation is unglamorous: inventory the tools in use, write down who reviews what, and make sure every teacher knows the rules. With that groundwork in place, a district can adjust calmly instead of reacting to each headline.