Most parents reading this are caught between two fears. Fear A: my kid will use AI to copy answers, learn nothing, and crash on the first real test. Fear B: I will hover so much that I make math feel awful, and my kid will hide the AI use anyway. Both fears are real. This guide is the middle path: a small set of supervision habits that catch the actual risks without making your kitchen feel like a courtroom.
The threat model
Cheating-by-AI is not the same as cheating-by-solution-manual. Solution manuals give an answer; AI explains steps. That makes copy-paste cheating both more useful in the short term and more obvious in the long term — because the missing skill shows up the next time the kid sees the topic without internet.
The two patterns to watch for:
- Pattern A — answer-only use. Kid types the question, copies the final answer, never reads the steps. Symptom: homework is done suspiciously fast.
- Pattern B — false confidence. Kid reads the AI explanation, thinks they understand, but cannot reproduce it on the next problem. Symptom: homework looks great, test scores drop.
The five-minute weekly check-in
Once a week, sit with your kid and run this exact script. Five minutes. No screens.
- "Show me one homework problem from this week and walk me through how you solved it." (No AI in the room.)
- "What is one thing you got wrong this week and what did you learn from it?" (Tests whether they used AI as feedback.)
- "Which topic from this week feels least clear?" (Surfaces gaps before the test.)
- "Anything you are stuck on right now?" (Re-opens the channel.)
- "Ok, what is your study plan for this weekend?" (Hands ownership back.)
Five questions, five minutes. That is the entire dashboard.
What to not do
- Do not read every AI conversation. It teaches the kid to use a different account.
- Do not test pop-quiz them on every topic. Math becomes adversarial.
- Do not punish wrong answers. Wrong answers are how learning shows up.
- Do not solve the problem for them when they get stuck. Ask "what would your first step be?" instead.
Concrete settings to enable
If your kid is using a shared computer or family device, a few low-touch settings help:
- Have one known AI tool the family uses — e.g., the AI-Math solver — bookmarked. Kids defaulting to a known tool reduces sketchy-app sprawl.
- Encourage paper first: have a stack of scratch paper near the computer. Visible paper changes behaviour.
- Set a homework window (e.g., 6:30–8:00 p.m.) and treat that time as quiet. Routines beat surveillance.
- Skip parental tracking apps unless there is a specific issue. They erode trust faster than they catch problems.
Signals that things are going well
- Your kid can explain why a step works, not just what it is.
- They sometimes correct the AI — "this method is faster than what it showed me."
- Test scores match or exceed homework scores.
- They mention math without being asked.
Signals to investigate
- Homework scores high, test scores low.
- Cannot redo a recent problem without re-opening the AI.
- "I just did what the AI said" when asked to explain.
- Refuses to show work.
If two or more of these appear for two weeks in a row, time for a calm conversation — not a confiscation.
How to talk about cheating without escalating
"I am not worried about you using AI — that is a normal study tool now. I am worried about how you are using it. Tonight, can you walk me through one problem with the AI tab closed? If you can do it, we are good. If not, we will figure out the gap together."
That sentence does three things: it normalises AI, it sets a real test of skill, and it offers help instead of punishment. Most kids will respond honestly to it.
The longer-term frame
Your kid is going to live and work with AI tools for the next 50 years. The skill we want to build is not "avoid AI" — it is "use AI well, recognise when it is wrong, learn from its explanations, and develop independent thinking on top of it." Math homework is the cheapest training ground for that skill that exists.