Family Conversations

Family AI rules

4 min read · Updated September 27, 2026

A family AI agreement works when it is short, specific and written together. Seven rules covering schoolwork, personal information, strangers, generated images, advice, checking facts and when to tell an adult will cover almost everything that actually comes up.

Most family technology rules fail for the same two reasons: there are too many of them, and they are too vague to follow. "Be sensible online" is not a rule, it is a hope.

What follows is a short framework you can adapt. Seven rules, each specific enough that a teenager could tell whether they had broken it.

Write it together

This matters more than the wording. A rule your teenager helped write is one they have to argue against themselves to break. A rule handed down is one they only have to get around.

It also surfaces things you did not know. Ask what they think the rule should be about schoolwork and you will learn a great deal about what is currently happening.

The seven rules

1. Schoolwork

AI can help you understand, quiz you, and give feedback on what you wrote. It does not write what you hand in. If you could not explain it, you do not submit it.

Check what the school actually allows first and make this match it. A household rule that contradicts the school puts your teenager in an impossible position.

2. Personal information

No full name, address, school name, phone number, passwords or financial details. No photographs with identifying detail in them. It feels private, but it is a message to a company.

3. Other people’s information

A friend’s situation is not yours to type in. No screenshots of other people’s messages, and no details about someone else’s family, health or relationship.

This one is frequently missing from family rules and comes up more than you would expect, because asking an AI for advice about a friend requires explaining the friend.

4. Generated images

No generating images of real people without their permission. Not classmates, not teachers, not as a joke. And treat any shocking image or clip you are sent as unverified until you have checked it.

Worth stating plainly rather than leaving to be inferred, because the social version of this spreads quickly and the harm is real.

5. Advice

AI is fine for ideas and explanations. It does not decide anything about health, money, safety, or a real relationship. For those, talk to a person.

6. Check what matters

If you are going to hand it in, act on it, or repeat it as fact, check it somewhere else first. It sounds equally confident whether it is right or making it up.

7. Tell an adult

If something it says frightens or upsets you, or if someone uses AI to target you, tell an adult. You will not be in trouble for telling, and you will not lose the device for it.

The part parents agree to

An agreement with obligations on only one side is not an agreement. It is worth writing down what you are committing to as well:

• I will tell you what I can and cannot see, accurately, and before I set anything up.
• I will not read your conversations. If something ever changed that, I would tell you first.
• I will not overreact when you tell me something difficult.
• We will revisit these rules as you get older, and they will loosen.

The last one does most of the work. A rule that visibly changes with age reads as care; a rule that never changes reads as distrust.

Adjusting by age

• 13 to 14: all seven rules, close involvement, devices out of bedrooms overnight, platform controls set up together.
• 15 to 16: the same rules, less checking, more explaining. Schoolwork and personal information are the ones worth holding firm on.
• 17 to 19: mostly advisory. Rules 4, 5 and 7 still matter; the rest is largely established habit by now, which is the argument for starting earlier.

What happens if a rule is broken

Decide this in advance and say it out loud, because a consequence invented in the moment always looks like temper.

Distinguish between the rules too. Handing in AI-written homework is a schoolwork problem with a schoolwork response. Generating an image of a classmate is a different category and should be treated as one.

Where TeenCue fits

TeenCue can tell you which supported AI services are being used and what kinds of topics come up, which gives an agreement like this something to be grounded in rather than assumed.

It cannot enforce any of these rules. It does not block services, it does not read conversations, and it will not tell you whether a piece of homework was written by a person. Rules are kept by agreement; this only helps you notice when a conversation is worth having.