1) Turn off “Improve the model for everyone”

Most AI apps have a setting that lets your chats be used to improve the tool. For a family account, I suggest switching this off.

In ChatGPT (quick path): Profile → Settings → Data Controls → toggle OFF “Improve the model for everyone.”
(Do this on phone or desktop—once it’s off, it applies to your account.)

2) Set a Family Password (your simplest scam shield)

Family Password Idea: Choose a secret word/phrase that only your family knows.

Use it when:

• Your child gets a strange message “from” a known person (WhatsApp/email/even a phone call).
• Someone says “your parents already know about this.”
• Anyone asks for help while sounding/looking like a parent/sibling/grandparent.

Rule (make it non-negotiable):
“If the person doesn’t know the family password, do NOT believe them and do NOT follow instructions.”

This one habit protects kids from social engineering—where someone manipulates your image or voice to scare them. 

3) Teach kids one sentence: “AI can hallucinate.”

AI can sound confident and still be wrong. It may invent facts, sources, names, or lists—without meaning to lie.

A simple rule that works:

• If it matters, verify it.  Ask AI to give you the original source – this could be an official website, a second trusted reference.)
• Never act on urgency. Pause, ask an adult, cross-check.

If you do just these three things, your child’s AI use becomes dramatically safer—without banning tools or creating fear.

If you would like to get your children to learn how to use AI safely for projects, research or just for fun take a look at our AI Training for children.