Use it for the scaffolding around your family, not for the parenting itself. Letters, lists, meal plans, the first draft of an awkward email: hand those over. Anything that needs to know your actual child, or that touches their health or their mood, does not go to a chatbot. And assume your child is watching which jobs you give away.
That last sentence is the part almost nobody tells you, and it is the reason this is worth ten minutes of thought rather than a list of prompts.
Your child is already ahead of you
Common Sense Media's research on AI attitudes among families, published on March 9, 2026, surveyed US children aged 12 to 17 and their parents. It found that under half of parents use AI regularly, against more than two thirds of kids and teens. The same research found that 84 percent of parents worry about AI misusing children's data.
Sit with those two findings together for a second. Most parents are worried about what these tools do with a child's information, and most children are using them more than their parents are. The worry is running ahead of the experience.
Meanwhile the Pew Research Center, in a report published on June 17, 2026 from a survey of 5,119 US adults fielded February 17 to 23, 2026, found that 49 percent of American adults have used a chatbot like ChatGPT, Gemini or Copilot, up from 33 percent in 2024. About a quarter use one daily.
So this is not a question of whether AI comes into your house. It is already in there, usually in your child's hands before yours. What you still get to decide is what it does for you, and what your child sees you refuse to let it do.
The Handover Test
Here is the rule I would use, and it takes about four seconds.
The Handover Test is the question to ask before you let AI do a job for you as a parent: if my child watched me do this, what would they learn about when a person should think? If the honest answer is "nothing important", hand it over. If it is "that you can skip the hard part", that job stays with you.
Rewriting a school letter into four dates on the calendar teaches a child nothing except that admin is admin. Hand it over. Asking a chatbot what to say to your daughter about her friendship group teaches her, if she ever finds out, that the hard conversations in this house get outsourced. Keep that one.
This is the family version of the principle I use with schools: let AI help with the doing, and keep the thinking with your child. Applied to yourself, it becomes: let it do your doing, and let your child see your thinking.
The jobs worth handing over
These are the ones that pass the test comfortably. They share a shape: the work is real but the judgment in them is small.
Turning school communications into dates. Paste in the newsletter, ask for every date, deadline and thing-to-bring as a list. Schools write in paragraphs. Families run on calendars.
Meals from what you actually have. "Here is what is in my fridge and cupboard. Give me four dinners for a family of five, nothing that takes over 30 minutes, one of them meat free." This is the use that surprises people most, because it solves a problem at exactly the hour nobody has energy left to solve it.
The first draft of the email you are dreading. Write the facts yourself in blunt, ugly sentences. Then ask for the tone to be softened, not the content changed. Read it. Change it back into your own words. Send yours.
Understanding something well enough to explain it yourself. If your child is learning something you last met at fourteen, ask the chatbot to explain it to you, not to them. That is a different request, and it ends with you at the table instead of a screen.
Ideas that use what you already own. "Six things to do with an eight year old on a wet Saturday, using only things most houses already have." It is better at this than it is at anything requiring taste, because the bar is low and the constraint is clear.
Plain English out of officialese. A school policy, a form, a letter about funding. Ask what it means and what, if anything, it asks you to do.
What I Would Do at Home
I use these tools every day, and I write a newsletter to more than 44,000 people about them, so I am not going to pretend to a purity I do not have. Two things I would still do, and they are habits rather than rules.
The first: I would do the checking out loud. When a chatbot tells you something and you go and confirm it, say so in front of your child, including the times you turn out to have been wrong. Children do not learn to doubt a machine by being told to doubt it. They learn it by watching an adult do it, which is a habit worth building deliberately.
The second: I would say out loud which jobs I am not giving it. Not as a lecture. Just as a remark, at the moment it comes up. "I could get it to write this, but it's your grandma's birthday, so no." Ten words. Your child will remember that longer than any family technology agreement pinned to a fridge.
The American Academy of Pediatrics puts the general version of this plainly in its family media plan guidance, updated January 20, 2026: "Remember, you are an important role model in practicing healthy digital habits and establishing a healthy relationship with technology." That is written about screens in general. It applies with more force here, because with AI the thing being modeled is not how long you look at a device. It is whether you think before you accept an answer.
Your child will not learn their AI habits from your rules. They will learn them from watching which jobs you were willing to give away.
The jobs I would keep
Short list, and I would be stubborn about it.
Anything that requires knowing your actual child. A chatbot has never met them. It will produce a fluent, reasonable, completely generic answer about "a 13 year old", and you will read it in the voice of someone who knows what they are talking about.
The apology. If you got it wrong with your child, the repair is the parenting. A well-worded apology that you did not word is not a repair.
Anything about their health, their eating, their sleep or their mood. Pew found that 20 percent of US adults have used a chatbot for medical advice. I understand the pull of that at eleven at night. It is still the wrong door. Your pediatrician, or your GP if you are in the UK, has seen a thousand children and yours in particular. If you are worried about your child's mental health, that goes to your doctor or to the school's safeguarding lead, the named member of staff responsible for child protection. A chatbot is not a clinician and does not know your child is your child.
The first conversation about something difficult. You can ask for help thinking afterward. Do not let it script the opening.
What it cannot promise you
Three honest limits, because the pages selling you prompt packs will not mention them.
It does not know when it is wrong. It produces the most likely-sounding next words, and a confident wrong answer looks identical to a confident right one. For anything that matters, check it against a source that has a name.
What you type in is kept by default. In ChatGPT, OpenAI's own data controls documentation describes a setting called "Improve the model for everyone", which is on unless you turn it off; with it off, "your conversations will still appear in your chat history but won't be used to train ChatGPT." There is also a Temporary Chat option, which OpenAI says is "deleted from our systems after 30 days" and is not used for training. If you are pasting in a school letter with your child's name, their class and their teacher's name on it, that is worth knowing before you paste. Strip the names. It does not need them to give you the dates.
The settings are not the safety. If your child uses these tools too, the account controls are a separate job from this one, and worth setting up properly.
What changes as your child gets older
Under 8. They do not reliably tell the difference between a machine that talks and a person who talks. Use it out of sight at this age, or if they are in the room, narrate it as a tool: "I'm asking the computer for dinner ideas." Do not let it become the thing that answers their questions. That gap between a child asking something and an adult answering is where a lot of the thinking happens.
8 to 12. They will copy your phrasing exactly. If you say "let me just ask AI," that is now the house reflex for not knowing something. Try "let me check" instead, and then actually check.
13 to 15. They will spot an inconsistency from another room. If you use AI for your work while their school bans it for homework, do not pretend the two are the same. Say why the line is where it is: your job is the output, theirs is the learning, so the machine can carry yours and not theirs.
16 to 18. Bring them in. Ask them to look at something a chatbot wrote for you and tell you what is wrong with it. It is the rare technology conversation where your teenager gets to be the expert, and it turns a rule into a shared habit.
What to say
You could say, when they catch you using it: "Yes, I use it. I use it for the boring half of things. Ask me what I don't use it for."
You could say, at the moment you refuse it: "I could get it to do this bit, but then I wouldn't have thought about it, and thinking about it is the point."
You could say, to a teenager: "What were the last three things you asked it? I'll show you mine."
One thing to do this week
Pick one recurring chore and hand it over completely: the school newsletter, the meal plan, the form nobody understands. Then name one job out loud, in front of your child, that you are not going to hand over. That is the whole practice. The first part gives you back twenty minutes. The second part is the one they will remember.
If your child's school has not yet spoken to parents about AI, that is usually a gap in staff confidence rather than a lack of care, and it is the work I do with schools: training teachers on AI in the classroom, setting a strategy, and making sure the policy means something by the time it reaches a classroom. It is worth sending them. If you want the wider picture first, the five things I teach every room of teachers is the place to start, and I write a newsletter that keeps teachers and parents up to date without the panic. Parents are very welcome.
Dan Fitzpatrick is a former teacher and assistant headteacher who now trains schools on AI in education. More about Dan.
