Did your child use AI for homework? How to tell and how to talk about it
Most reliably by whether your child can explain the text in their own words. Homework written with AI tends to be too polished for the child's age, generic and free of the mistakes the child usually makes. A detector can point to suspicious passages, but short school texts are not reliable material for it and the result proves nothing — no AI text detector is a hundred percent accurate. Rather than checking and punishing, an agreement works better: what AI may be used for, what the child has to write themselves and how the school sees it.
What to notice
A parent knows how their child writes better than any tool. A text that does not match the child's age is suspicious: long sentences, words the child does not use and none of the mistakes they usually make. So is homework finished in a few minutes when the same kind of work used to take an hour.
The second trace is the content. Generated homework answers the topic in general terms, does not connect to what was covered at school and lacks the specific experience the assignment asked for. With a "what I did over the summer" essay you notice it fastest — AI writes a nice text about the summer holidays, just not about yours.
Why a detector is not enough
A detector needs continuous text of a reasonable length. Homework often has only a few sentences, and there the tool gets it wrong more often than with a longer piece. Younger children also write simply and predictably, which is exactly what detectors look for in machine text, so even homework written independently can come out as suspicious.
Even when a detector marks passages, it is only a pointer to where to ask. A child who wrote the homework themselves should not be reproached because of a number. Going through the text together tells you far more: what they meant by it, where they know it from and how they would put it differently.
How to agree on rules
A ban on AI at home is hard to enforce, and your child will come across it anyway. It is more useful to agree on what they may use it for: having material explained, checking an answer or practising exercises. The text that gets handed in, though, they should write themselves. Align the line with what the school says — rules differ between schools and between teachers.
If you find out that AI wrote the homework, treat it as a chance for a conversation, not as an offence. Let your child rewrite the homework their own way and ask why they did it — it may be a lack of time, uncertainty or fear of making mistakes. If the school ends up dealing with the homework, it is better for them to hear about the use of AI from the child themselves.
FAQ
Should I check homework with a detector before it is handed in?
Doing it routinely makes little sense; short children's texts are not reliable material for a detector. It is better to have your child retell the homework — if they can explain it, it is theirs.
Will the teacher be able to tell?
Often yes, because they know how the child writes at school. The difference between homework done at home and work done in class tends to show even without a tool.
Is help from AI with homework cheating?
It depends on what the child used it for and what the school allows. Having material explained is usually not a problem; handing in text from AI is.
Try it on your own text
Paste a text or upload a file and look at the score and at the specific sentences that came out suspicious.
Run a detection