How to tell if AI wrote your kid's homework (without accusing them)
Detectors are unreliable. The two things that actually work are reading for specific signals in the text — flawless structure, no personal detail, vocabulary above their level, tidy generic examples — and asking your child to explain their reasoning out loud. If they can walk you through why, the thinking is theirs.
The question almost never arrives calmly. You read a paragraph your child wrote, and something about it does not sound like them. Cleaner. Flatter. Older. And then you are stuck holding a suspicion you cannot prove and do not want to say out loud.
The signals that actually mean something
AI writing has a texture. Once you know what to look for, you stop guessing:
- Every paragraph is the same length and follows the same shape. Real kid writing is lumpy.
- No specific detail. No name of their teacher, their book, their experiment, their own opinion. AI does not know those things unless it was told.
- Vocabulary that outruns their speech. If your child has never said "furthermore" out loud, it is worth a question.
- Examples that are tidy and generic — "a student named Sarah" rather than something from their actual week.
- A confident conclusion that restates the intro without adding anything.
The signals that mean nothing
Plenty of honest work looks suspicious. Neurodivergent writers, ESL students, and kids who write formally on purpose all get flagged constantly. So does anything that follows a template the teacher handed out. And AI detectors themselves are unreliable enough that a percentage score should never be the thing you act on.
The two-minute conversation that settles it
Reading the text tells you whether to ask. The asking is what gives you an answer. Three questions, asked casually, at the table:
- "Walk me through how you got to this part." Understanding survives paraphrasing. Pasted text does not.
- "What would change if this number — or this fact — were different?" This is the one that separates memorised from understood.
- "Which bit of this did you find hardest?" A kid who did the work always has an answer. A kid who did not usually says "none of it."
Notice none of those are accusations. You are not asking whether they used AI. You are asking what they understood, which is the thing you actually care about.
What to do if the answer is yes
Before the conversation
Let it read the work first.
Paste the paragraph or snap a photo. You'll get what looks like your kid, what doesn't, and the questions to ask out loud.
Using AI is not automatically cheating — it depends entirely on the assignment. Our breakdown of where the line sits by subject covers what counts as help versus a shortcut for essays, math, coding and research. The productive move is almost never a punishment. It is redoing one paragraph out loud, together, so the understanding actually lands.
If you would rather not do the reading yourself, paste the work below and we will point out what looks like your kid and what does not — plus the questions to ask.
Frequently asked questions
How can I tell if my child used AI to write their homework?
Read for four signals: uniform paragraph structure, no specific personal or classroom detail, vocabulary above your child's spoken level, and generic examples. Then ask them to explain their reasoning out loud — understanding survives paraphrasing, pasted text does not.
Are AI detectors reliable for checking my kid's homework?
No. AI detectors produce frequent false positives, especially for neurodivergent writers, English language learners, and any work that follows a teacher-provided template. Treat a detector score as a prompt to have a conversation, never as evidence.
What should I say to my kid if I think AI wrote it?
Skip the accusation and ask about the work instead: walk me through how you got here, what would change if this number were different, and which part was hardest. Those three questions reveal understanding without putting your child on trial.
Is it cheating if my child used AI for homework?
It depends on the assignment. Using AI to explain a concept, check spelling, or generate practice questions is usually fine; having it produce the submitted work is usually not. Ask the teacher what they allow for that subject, or use a written family rule so the line is clear in advance.
Have a specific piece of homework in mind?
Paste it — or a screenshot of it — into the Homework Check on our homepage and ask your question. You'll get a plain-language read on whether the thinking looks like your kid's, what stood out, and the questions worth asking them. First check is free.