Five AI house rules worth putting on your fridge
Families who write down their AI rules and put them somewhere visible get followed rules. Families who keep the rules in their head do not. Here are the five that come up in almost every session.
Two families in the same workshop had, by their own account, near-identical views on what their kids should and shouldn't do with AI. One family had those views written on an index card stuck to the fridge. The other kept them "understood." A month later, the fridge family's rules were still being followed. The other family's eleven-year-old had, without meaning to break any rule exactly, pasted a full page of his sister's diary into a chat to "get writing tips" - because no one had ever actually said, out loud, that this counted as the kind of thing the family didn't do.
Rules that live only in a parent's head are not rules a child can follow, because a child cannot check something they were never shown. That is the entire case for writing them down, and for putting them somewhere visible rather than somewhere remembered.
What AI rules should families have?
The five below are the ones that come up, in some form, in almost every family workshop we run - not as a complete list, but as a solid starting point any family can adapt.
- Never type in your full name, address, school name, or a password.
- Never let it read something private that belongs to someone else - a sibling's diary, a parent's messages - without asking first.
- Always check anything specific or personal against a real source before believing it.
- Anything that leaves the house - sent to a teacher, posted publicly, shared with a stranger - needs a grown-up's yes first.
- If something it says feels wrong or confusing, stop and ask a person, not the AI, what to do next.
Why visible beats remembered
A rule that has to be recalled under pressure - mid-chat, mid-curiosity, mid-"just this once" - competes against the moment's excitement and usually loses. A rule stuck on the fridge, read that morning over breakfast, is already loaded before the moment arrives. Visibility is not a nicety here. It is most of what makes a rule actually work.
Rules that are visible are rules that get followed. Rules that are merely understood are rules that get reinterpreted.
Sorting jobs into red, yellow and green
One honest limit: five rules cannot anticipate everything a curious child will eventually try. What they can do is establish the habit of checking before acting in the categories that matter most, which is worth far more than a longer list nobody remembers. The next article in this level covers a specific, sharper danger these general rules don't fully cover on their own - a hidden instruction sitting inside something the AI reads.
Questions we get asked
What AI rules should families have for kids?
A solid starting five: never share your name, address, school or a password; never let it read someone else's private information without asking first; always check specific or personal claims against a real source; get a grown-up's yes before anything leaves the house to a teacher, a stranger, or the public; and stop and ask a person, not the AI, if something feels wrong.
Why do family AI rules need to be written down instead of just discussed?
A rule that only lives in memory has to be recalled in the moment, competing against curiosity or excitement, and it usually loses that competition. A rule written down somewhere visible - like a fridge - is already front of mind before the moment arrives, which is most of what makes it actually get followed.
How do you decide which AI tasks a child can do alone versus with a parent?
Sort specific tasks - not vague categories - into three piles as a family: fine to do alone, needs a grown-up's involvement first, and never. Doing this sorting together, out loud, and revisiting it occasionally works better than a single blanket rule that tries to cover every situation at once.