Most of the AI in your child's school sits on the teacher's side of the desk: the lesson plan, the quiz, the letter home and sometimes the marking. Pupils meet it far less, and the bans in the news stop pupils, not teachers. How much touches your child depends on the school, and in most places nobody has to tell you. Asking is the only way to know.
Here is where to look, what "we have an AI policy" usually means, and the words to use with the teacher and with your child.
Where AI actually is in a school
Think of four places, and notice which one the news is about.
The planning desk. This is where most of it lives. In Gallup's May 2026 survey of 2,069 US public school teachers, run with the Walton Family Foundation in February and March 2026, 60% said they use AI for their work and 30% use it at least weekly. Gallup's June 2025 survey of the same teacher panel broke that down: 37% used AI at least monthly to prepare lessons, 33% to make worksheets and activities, 25% to write quizzes and assessments. In England, Ofsted's June 2025 study of 21 early-adopting schools and colleges found the same shape: teachers planning lessons, adapting resources and creating quizzes.
The office. Ofsted found administrative staff using AI to write letters to parents, summarise long documents and update policies. The letter you were sent about the new uniform may have had a machine's first draft. That is the doing, not the thinking, and I have no quarrel with it.
Marking and feedback. The least common use, and the one parents ask me about most. Gallup's June 2025 figure was 18% of US teachers using AI at least monthly for grading and feedback. Ofsted's early adopters were, in its words, more cautious about assessment than about planning, worried about accuracy and about teachers becoming less aware of their own students.
Your child's screen. The smallest of the four in most schools, and the only one the headlines are about. Ofsted found that most primary schools in its sample had not allowed independent pupil use at all, and that secondary schools tended to model AI in front of a class before letting students touch it. Where pupils do have a tool, it is usually inside a school account (Google's Gemini switched on in Classroom, for instance, which I covered in the Gemini post), and the school chose it, or a default did.
So when a parent asks me "is my child's school using AI?", the honest answer is nearly always yes, and nearly always in the three places the policy does not mention.
Is a teacher using AI to mark my child's work?
Possibly, and it is the use with the least guidance around it. In Gallup's May 2026 survey, 58% of US teachers said they had received no guidance at all on using AI for grading or feedback, formal or informal; only 8% had formal guidance. An NBC New York investigation published on 31 August 2026 found teachers using AI for lesson plans, exams, grading and report card comments, and one Connecticut district, Amity, rewriting its policy so that students are told when AI has evaluated their work and AI "may not serve as the sole basis" for a grade.
That Amity line is what good looks like. The risk is not that a machine read your child's essay. The risk is that nobody did. A teacher who uses AI for a first pass and then reads every piece, decides every grade and writes the comment in their own words is using the tool the way I would want. A grade that goes from the machine to the mark book with no human in between is a different thing, and a school should be able to tell you which of the two it does.
What does "we have an AI policy" actually mean?
Usually a rule about pupils. In RAND's September 2025 report, drawn from its spring 2025 surveys of US teachers, principals and district leaders, 45% of principals said their school or district had a policy or guidance on AI, and 34% of teachers said they had a policy on AI and academic integrity, which is the "don't cheat with it" clause. Gallup's May 2026 survey found only 18% of teachers had formal guidance from their administrators on their own use of AI, and 34% had none of any kind.
Put those together and you have the typical school: a policy that tells pupils what they may not do, and silence on what the adults are doing. Nobody is hiding anything. Nobody was asked.
The early adopters look different. Ofsted found schools updating their AI policies at least termly, folding AI into existing safeguarding and data protection policies rather than writing a separate one, and in some cases running an approval committee that checks a tool for data compliance and for whether it helps learning before anyone is allowed to switch it on. That last detail is the one to listen for.
In England, the Department for Education's guidance on generative AI in education, last updated 12 August 2025, asks schools to be clear about when AI is acceptable for educators and for pupils, to review homework policies, to comply with the age limits of any tool, and to keep personal data out of AI unless strictly necessary. On parents, it says schools "may also wish to consider how they engage with parents". May wish. Not must. In the US there is no federal requirement that a school tell parents; the National Parents Union survey of 1,511 public school parents (12 to 18 February 2026) found 57% said their child's school had not asked them for input on AI.
That gap is the thing to understand. In most places, a school's duty to tell you what it does with AI is the same as its duty to tell you which textbook it bought: it exists in good practice, not in law. Which is why the question has to come from you.
What about the bans in New York and Los Angeles?
They stop the fourth place and leave the first three alone. New York City's moratorium, announced on 2 September 2026, pauses student-facing generative AI from 2-K to 8th grade for the 2026 to 2027 school year, and the same document says teachers may continue using AI for planning and operational tasks. Los Angeles Unified, the second largest district in the country (a district is the US equivalent of a local authority or an academy trust), paused all student use of generative AI on district devices "until further notice" on 2 September, and Education Week reports that the moratorium does not apply to teacher use. LAist reported that board members and parents in the room learned of it at the same committee meeting; one board member's reaction was that the word needed to get out.
Both are decisions about the pupil's screen. Neither changes whether the worksheet was drafted by a machine or whether the essay was read by one. If your child is in one of those districts, the Kitchen Table Question still applies: what does this change about what my child meets at home tonight? In most homes, nothing.
What I See in Schools
The pattern I see most often is that AI arrived through the staffroom door months before it reached the policy. A handful of teachers started using it on their own accounts to plan lessons and write reports, a few more copied them, and by the time the school wrote anything down the writing was about pupils and cheating, because that was the question parents and governors were asking. Staff use stayed a private habit. When I trained teachers in my years as a Director of Digital Strategy in further education, and in the sessions I run with schools now, the question I hear most from staff is not "should we use this?" but "are we allowed to?", and most of them do not know the answer because nobody has written one.
The second pattern is at parents' evening. A parent asks "do you use AI?" and hears "we have a policy". Both people leave the table satisfied and neither has learned anything. The policy is real. It is also, more often than not, a rule about homework.
Ask where AI touches your child's work, and who checks it. "Does the school use AI?" only has a policy for an answer. "Where?" has a person.
What to ask the school
One question, then three follow-ups. The one question is what I call the Where Question: "Where does AI touch my child's work, and who checks it?" It works because it cannot be answered with a policy. It has to be answered with a list.
The follow-ups, in order of how much they matter:
- Which AI tools have pupils been given inside their school account, and who decided? A good answer names the tools and the person or committee. "It came switched on" is an honest answer too, and a useful one.
- If AI helps with marking, feedback or reports, does a teacher read every piece before a grade or a comment reaches my child? A good answer is yes, with a description of how. The Amity wording above is the standard to hold in your head.
- What information about my child goes into any AI tool, and has somebody checked where it goes? In England the answer should reference data protection and the school's own privacy notice, which the DfE guidance requires. In the US, ask whether the tool has been through the district's approval process and signed a student data privacy agreement with the district.
Send it as an email rather than saving it for parents' evening; the teacher may need to ask someone. You could write:
"I am not worried, and I am not asking the school to stop anything. I would just like to understand where AI is being used in [name]'s learning this year: which tools pupils have, whether it helps with marking, and who checks it. Could you point me to the person who knows, or to the part of the policy that covers staff use as well as pupil use?"
In the UK the person is usually a deputy head or the designated safeguarding lead (the member of staff responsible for child protection), because the 2025 edition of Keeping Children Safe in Education, the statutory safeguarding guidance for schools in England, brought AI inside its guidance on schools' filtering and monitoring duties. In the US it is more often a principal or the district's instructional technology lead.
What to say to your child, by age
Under 8. Nothing about AI at school unless they raise it. What matters at this age is the teacher, and Ofsted's finding that most primary schools have not let pupils use AI independently is the most likely state of your child's classroom. You could say, if a worksheet looks oddly generic: "Who made this sheet, do you think? Your teacher, or a computer helping your teacher?" Then leave it. You are planting the idea that adults use tools, nothing more.
8 to 12. Start with feedback, because that is where a child first notices. You could say: "When your work comes back with comments, do you think a person wrote them? How can you tell?" If the answer is "I don't know", that is the point; use the three checking questions (Who says? How could we check? What would it look like if this were wrong?) and check together.
13 to 15. Ask what is in their school account, because they know and you do not. You could say: "Show me what's in your school Google or Microsoft account. Is there an AI in there? Are you allowed to use it for homework?" Expect the fairness question to come back at you: "The teachers use it, so why can't we?" The honest answer is that the teacher has already learned the thing the machine is helping with, and your child has not yet, which is the whole of the Homework Test: if the machine did this part, what would my child lose?
16 to 18. Find the school's policy together and read it, then find the exam board's rule, which is the one that matters for coursework and can end a qualification. You could say: "Let's find out what the school actually says, and what the exam board says, before you assume either." At this age the school's rule on what pupils may do is the rule your child is examined under; what you allow at home does not change it.
What none of this can guarantee
A policy tells you what a school intends, not what happens in every classroom. No filter or setting shows you what a teacher typed into a tool on their own phone; the only route to that is the conversation above. A tool the school vetted in September may change its terms by January, which is why Ofsted's early adopters review policies every term and why I would ask the Where Question every year, not once. And a straight answer depends on the school having asked itself the question; some have not yet, and your email may be the first time anyone did.
Where the rules stack up: the school's rule on pupil use applies at school whatever you allow at home; the exam board's rule wins on coursework; the platform's age limit applies to any tool your child logs into; and anything that worries you about what a child was shown or told goes to the safeguarding lead (in the US, the principal or counsellor), not to a settings page.
Something to do today
Send the email. One paragraph, no accusation, one question and three follow-ups. If you want the fuller checklist, that is the next post in this series; for tonight, the Where Question is enough, and it is more than most parents have asked.
If your child's school has not spoken to parents about AI yet, this is the work I do with schools: helping a school build an AI strategy that covers staff use as well as pupil use, and that a parent can read and understand. It is worth sending to your child's head of year. And I write a Sunday newsletter that keeps teachers and parents up to date without the panic; parents are very welcome.
Dan Fitzpatrick is a former assistant headteacher and Director of Digital Strategy, has trained more than 150,000 educators, and writes about AI in education for Forbes. More about Dan.
