Two family AI guides trust a kitchen-table check, not a setting

The classroom note

A parent sitting with a child over a maths worksheet that the child has already run past a chatbot faces a specific, checkable question, not an abstract one about AI in general: did the answer come from understanding, or from copying? Two family-facing guides, from different sides of the Atlantic, converge on the same kitchen-table method rather than a blanket rule. Internet Matters' family conversation guide, aimed at parents of children from early years to teens, and Common Sense Media's Youth AI Safety Institute review of ChatGPT-5, aimed at families already letting a teenager use it, both treat homework help as a legitimate use that still needs a check, rather than an activity to ban outright.

What the evidence says

The two documents are different genres and should be read as such: Internet Matters' guide is a practical, age-banded conversation tool, for tweens it suggests a ‘research challenge’ of asking AI a question and then checking a textbook or trusted website, rather than a study of outcomes. Common Sense Media's document is a structured product review, rating ChatGPT-5 ‘High risk’ overall while naming homework help and Study Mode as legitimate uses, distinct from its central concern, teen use of the same chatbot for mental-health or emotional support. Neither source measures whether these strategies change a child's schoolwork; both are guidance for a family to use, not evidence that using them works.

The implementation question

What a family can actually verify at home is narrower than ‘is AI good or bad for homework.’ Common Sense Media's review notes that parental controls on ChatGPT exist but only function if a parent sets them up and a teenager consents to linking the account, and that a determined teenager can use an unlinked account to bypass them entirely, a limit on any technical control, not just this one. Internet Matters' guide sidesteps controls altogether and asks a parent to do something checkable regardless of settings: read the AI-generated answer together, ask where a fact came from, and compare it against a second source before treating it as homework help rather than borrowed writing.

What holds and what fails

What holds, across both, is that verification is a task for a person, not a setting: ‘does this match what a book says’ and ‘did you write parts of this yourself’ are questions a parent can ask without any technical knowledge. What fails is trusting a parental control alone to enforce a boundary a determined teenager can route around, which both sources acknowledge in different ways, Common Sense Media directly, Internet Matters by building its guide around conversation rather than restriction.

• Can your child explain, in their own words, an answer AI gave them for schoolwork?
• Have you asked where an AI-cited fact actually comes from, together?
• Do your household's rules rely on a setting, a conversation, or both?

Neither guide promises that talking at the kitchen table will catch every misuse; both are built on the more modest claim that a check anyone can do beats a rule no one enforces.