Ask your child's school ten things: whether a written AI policy exists, where AI touches your child's work, who approved the tools, what happens to student data, who checks AI-assisted grading, what counts as cheating, and what happens if something goes wrong. A good school can answer in plain words. What it must tell you depends on your country and state, and the school's own rules come first.
Why ask now?
Because "does the school have a policy?" no longer has a safe default answer of yes. On October 13, 2026, New York Focus reported on a crawl of 678 New York school districts' public websites, with a reporter checking the results by hand. About 270 districts had a formal AI policy. Roughly 400 had none the reporters could find online, and more than 70 of those confirmed they had no board policy at all; 39 of the 70 said they plan to write one this school year. That is one state, the figures are approximate, and some districts have staff guidelines that never reached a website. It is still a useful reminder that silence is common.
Meanwhile, children are using the tools. The Center for Democracy and Technology (CDT) surveyed US teachers and students about the 2024-25 school year. As cited in its June 24, 2026 testimony to the New York City Council, 85% of teachers and 86% of students said they had used AI, and fewer than half of the teachers and students said they had received training or information about using it in class. CDT's October 2025 report also found that more AI use in a school went along with more risks to students, among them data breaches and tech-enabled sexual harassment and bullying. That is an association. It does not show that AI caused any one incident.
And families disagree with each other. Common Sense Media's March 9, 2026 research, done with Echelon Insights and Lake Research Partners among kids ages 12 to 17 and their parents, found that most kids and teens think using AI on school assignments is innovative and should be encouraged, while most parents think it is unacceptable and should carry consequences. The same assignment can be fine at one kitchen table and cheating at the next. The school's answer is what settles it, which is why it is worth getting.
How should I ask?
In writing, kindly, and two or three questions at a time. A teacher often has to check with someone else, and an email gives them time to do that. A parent-teacher conference (parents' evening in the UK) is a poor place for a question that needs a name and a date.
If you only ask one thing, make it the one 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.
You could write:
"I'm not worried, and I'm not asking the school to stop anything. I'd like to understand how AI is used in [name]'s learning this year. Could you point me to the person who can answer a few questions, or to the policy that covers it?"
The questions below go roughly from the most basic to the most serious. Each one says why it matters and what a good answer sounds like. A copy-and-send version is near the end.
The ten questions
I read the sources below on October 17, 2026. School and government pages change, so check the date on anything you rely on.
1. Is there a written AI policy, and can I read it?
A policy is the first thing to ask for because everything else hangs on it. A good answer is yes, here is the link, and here is the date it was last reviewed and who owns it. "We are working on it" is an honest answer too; ask when. The New York Focus review found that policies, where they exist, mostly set broad principles such as academic integrity rather than saying which uses are allowed. That is not wrong, but it means you should follow up with question 2.
2. Where does AI touch my child's work, and who checks it?
The Where Question. A good answer is a short list: the tools students have been given inside their school accounts, the tasks teachers use AI for, and the person who is responsible for each. If the answer is "I'll find out", that is also fine; ask when. For the full version of this conversation, see how AI is actually used in schools, from the inside.
3. Who approved each tool, and how?
A tool a teacher found last week and a tool the school has vetted are different things. A good answer names a person or a committee and describes a check on two counts: whether the tool handles student data properly, and whether it helps learning. England's Department for Education (DfE) product safety standards, updated January 19, 2026, are written mainly for suppliers, but schools can use them as a benchmark. They say, for example, that products should not give final answers or complete worked examples by default, and that they should avoid trying to build a personal relationship with the learner. In the US, the Future of Privacy Forum publishes a checklist for vetting generative AI tools (published April 24, 2024; the page shows an update on June 27, 2026). You do not need to read either. You need to hear that something like them was used.
4. What happens to my child's data?
Ask three smaller things: what the tool collects, whether anything a child types is used to train the company's AI model, and how long it is kept. The FPF checklist lists the training question as one schools should check. The DfE standards say products should not collect, store, share or use personal data for commercial purposes unless an appropriate lawful basis is confirmed. A good answer is specific: a named agreement, a privacy notice you can read, a yes or no on training. "It's secure" is not an answer.
In the US, the Family Educational Rights and Privacy Act (FERPA) gives you rights over your child's education records. According to the Department of Education's FERPA page, you can inspect them, the school must respond within a reasonable time and no later than 45 days after you ask, and it must answer reasonable requests for explanations of them. You can also ask for corrections, and the school has to tell you each year what your rights are. Whether a particular chat log counts as an education record can depend on who holds it and how, so ask in writing rather than assume.
5. What can the school see of my child's AI use, and who reads the alerts?
Many schools run monitoring software on school accounts and school devices. The useful answers are which product, what it covers, how long the school keeps what it collects, and who looks at a flag and what happens next. What GoGuardian and Gaggle actually see sets out what two of the best-known products' own documentation says, and the short version is that it depends on settings your district chose. For personal accounts at home, see what the school can see on a teen's personal account.
6. Does AI grade or comment on my child's work, and does a teacher make the final call?
A good answer is that a teacher reads every piece, decides every grade and writes the comment in their own words, even if a tool helped with a first pass. A grade that goes from a machine to the grade book with no person in between is a different thing, and a school should be able to say which it does. I went through the numbers in do teachers use AI to grade.
7. What counts as acceptable AI help on homework, and what happens before anyone is accused?
Rules that say only "do not cheat" leave a twelve year old guessing. A good answer says what is allowed by subject and age, whether students must say when they used AI, and what the school does first when it suspects misuse. Ask in particular whether a detector score can ever decide a case by itself; a good answer is no, and it describes other evidence, such as drafts and a conversation with the student. If a child has already been accused, start with what to do when your child is accused of using AI. For the wider question of where help ends, see is using AI for homework cheating.
8. What are students taught about AI, and what training have staff had?
CDT's figures suggest many schools use AI before they train anyone to use it. A good answer says what students learn (how these tools get things wrong, when to check, when to stop) and describes some staff training, even if it is basic. "Teachers have access to resources" is thinner than a session with a date.
9. What happens if an AI-made image of a student circulates, or a chatbot conversation upsets a child?
This is the question to have answered before you need it. A good answer names the person responsible (in England, the designated safeguarding lead; in the US, usually a counselor or administrator), describes what happens in the first hour, and says how both a targeted child and a child who made or shared something are supported. CDT's testimony reported that fewer than a quarter of teachers said their school had clear policies for responding to deepfake incidents involving students.
In England, the new edition of the statutory guidance, Keeping Children Safe in Education 2026, applies from September 1, 2026. According to the SWGfL summary, it now covers harmful interactions with generative AI that simulates human contact, such as chatbots and companion-style systems, and it treats AI-generated and digitally altered sexual images as part of its nudes and semi-nudes guidance. A parent in England can ask how the school applies that. I did not find a national US equivalent, and state rules vary, so a parent in the US should ask the district what applies.
If your child seems distressed by something they saw or said to a chatbot, talk to the school's safeguarding lead or counselor and to your pediatrician or GP. In the US you can call or text 988; in the UK, Samaritans is on 116 123 and Childline on 0800 1111. A chatbot is not a clinician.
10. Can I opt my child out of anything, and how will I hear when this changes?
Two questions, one answer you need: who tells parents, and when. A good answer gives a review date, a way to hear about a new tool before it is switched on, and a plain account of what an opt-out would mean for your child in practice. I wrote up the details in can I opt my child out of AI at school.
The ten at a glance
Save this table or print it. The last column is when to ask again.
| # | Ask | A good answer sounds like | Ask again if |
|---|---|---|---|
| 1 | Is there a written AI policy? | A link, a review date, an owner | It only covers cheating |
| 2 | Where does AI touch my child's work, and who checks it? | A short list with names | You hear "we have a policy" |
| 3 | Who approved each tool? | A person or committee, and two checks | "A teacher found it" |
| 4 | What happens to my child's data? | A named agreement; a yes or no on training | "It's secure" |
| 5 | What can the school see? | A product, a scope, a retention time | Nobody knows who reads alerts |
| 6 | Does AI grade my child's work? | A teacher reads and decides | The grade comes from a tool alone |
| 7 | What is acceptable help? | Rules by subject and age | A detector score settles it |
| 8 | What are students and staff taught? | Something with a date | "Resources are available" |
| 9 | What if something goes wrong? | A named person and a first-hour plan | No one is named |
| 10 | Can I opt out, and how will I hear? | A review date and a route | There is no route |
What I See in Schools
The pattern I see most often is that the person who can answer is not the person who gets the email. A parent writes to a class teacher, the teacher is not sure, and the question waits for a deputy head or a data protection lead who never saw it. In the AI strategy work I do with schools and trusts, and from my years as a Director of Digital Strategy in further education, the schools that handle this well decide in advance who owns the reply. That is a small thing, and it is the difference between a parent hearing back in two days and not hearing back at all.
The second pattern is that the first answer to a hard question is often thin, and the second is much better. Schools tend to open with the policy because the policy is what they have. Asking again, politely, with the Where Question, is usually what produces the list.
A good answer to a parent's question about AI is a name and a date, not a policy number.
What should I say to my child?
Tell them what you asked and why, so it does not feel like surveillance.
Under 8. Keep it light. You could say: "I'm asking your teacher how the computer helps in class, so I know how to help you at home."
8 to 12. You could say: "I asked school what AI you're allowed to use, so we both know the rules. You're not in trouble."
13 to 15. You could say: "I asked what the school can see on your school account, so I can tell you instead of guessing. Your own phone is a different thing, and I'd like to talk about that too."
16 to 18. Treat them as nearly adult. You could say: "Here is what I asked and what they said. Which of their rules do you think are unfair?" At this age, their view is useful evidence.
What can these questions not tell me?
Less than you would like. A policy on paper does not prove practice. A school can only answer for the tools it knows about, and children often use personal accounts it has never seen. An approved tool is not a safe tool; the DfE standards are written mainly for suppliers, and meeting them is a supplier's claim until someone checks. FERPA is about records and consent, and the page I read sets out no AI-specific duties. The school's published policy, your state or country's law, and a professional's advice if your child is struggling all come before anything in this article.
Something to do tonight
Pick the three questions that worry you most, put them in one short email, and send it to the class teacher or the school office. Add the Where Question if it is not one of the three. Then write the date in your calendar for a week from now.
The reason I wrote this one is the school itself. If your child's school has not yet spoken to parents about AI, this is the work I do with schools, and my guide to writing a school AI policy is worth forwarding to the principal or head teacher. I also write a weekly newsletter that keeps teachers and parents up to date without the panic. Parents are very welcome.
Dan Fitzpatrick is a former secondary teacher, assistant headteacher and Director of Digital Strategy who now trains schools on AI. More about Dan.
