How to stop your kid from cheating on homework with AI: a practical guide for parents

Homework help used to mean a parent, a tutor, or a friend on the phone. Now it is a text box that answers anything in seconds, never gets tired, and never says “figure it out yourself.” Pew Research Center reported in early 2026 that 54 percent of U.S. teens have used AI chatbots for help with schoolwork, about three in ten use a chatbot every day, and one in ten say they do all or most of their schoolwork with one.

The gap between what kids do and what parents know is the part that deserves your attention. Pew found that 64 percent of teens say they use chatbots, while only 51 percent of parents say their teen does, and about three in ten parents are not sure either way. Common Sense Media’s survey found the same blind spot: among teens who had used generative AI, only 37 percent of their parents knew. And roughly four in ten parents in the Pew survey had never talked with their teen about chatbots at all.

It also helps to know where kids themselves draw the line. When Pew asked teens what counts as acceptable, 54 percent said using ChatGPT to research a new topic is fine, 29 percent said the same about solving math problems, and only 18 percent said it is acceptable for writing essays. Most kids know where the line is. The problem is that the tool sits on the other side of it, one tab away, at nine o’clock the night before something is due.

Not every use of AI is cheating, and pretending otherwise will lose you the argument in about thirty seconds. Explaining a concept a different way, quizzing your kid before a test, or checking whether a sentence is grammatical is the kind of help a good tutor gives. Writing the paragraph, solving the problem set, or producing the answer the teacher will grade is not.

That distinction is not only about honesty. It is about whether any learning happened. A study of nearly a thousand high school math students, run by researchers at the University of Pennsylvania and published in PNAS, gave one group access to a ChatGPT-style assistant while they practiced. Their practice scores jumped 48 percent. Then the assistant was taken away for the exam, and the same students scored 17 percent worse than classmates who had never used it. A version of the tool built to give hints instead of answers erased the harm, though it did not help either. Most telling of all: the students who copied answers did not notice they had learned less.

You do not need an AI detector, and you should not trust one. Detectors miss most AI-written text and wrongly flag honest writers — especially kids still learning English — and our guide for teachers covers why in detail. The signals a parent can actually see are more ordinary, and more reliable.

The worst version of this is the ambush: a printed essay, a raised voice, and a kid who shuts down. The best version happens before anything has gone wrong, and it is shorter than you think.

Then make the rules easy to follow. That is what the next section is for, because a rule that depends entirely on a fifteen-year-old’s willpower at nine o’clock at night is not much of a rule.

These work best together. Each one closes a gap the others leave open, and none of them requires reading your kid’s messages.

Nullgen is a browser extension for the computer your child does homework on. It recognizes AI prompt boxes on any website — ChatGPT, Gemini, Claude, Copilot, and the new ones that launch every week — and stops typing and pasting into them. The rest of the page keeps working, so research, Google Classroom, the library catalog, and video lessons are exactly as they were. Only the prompt is blocked.

It is not surveillance. Detection runs on the computer, and nothing your child types is recorded or sent anywhere — not to Nullgen, and not to you. You are not reading their messages or watching their screen. You are removing one shortcut, the same way you would not leave the answer key on the table.

The free extension is enough for one computer, and installing it takes about a minute. The Family plan is for the rest of the house.

What changes in practice is small and immediate. The homework computer stops offering an easy way out, your kid stops having to decide not to take it, and the conversation you had in the last section stops depending on memory.

Honesty first: today, Nullgen runs in desktop browsers — Chrome, Edge, Firefox, and Safari on a Mac. It does not run on a phone yet, and no browser extension can reach inside a phone app. If your kid opens ChatGPT on their phone, the homework computer being protected does not change that.

In the meantime, the house rules carry the weight. The homework computer is where the assignment lives — the document, the portal, the submission — so using AI on a phone means retyping everything on the computer, which is friction most kids will not bother with. Keep the phone in the kitchen during homework hours, use the parental controls you already have (Apple’s Screen Time and Google’s Family Link can limit or block individual apps), and save your energy for the conversation. No tool replaces that, and this one is not trying to.