Put together a family safety kit. Return JSON {"rules":[...],"parent_settings":[...],"talk_script":"..."}: rules - simple rules for a child; parent_settings — settings from the parent’s side; talk_script - how to kindly explain this to a child. Return only JSON, no text around it.
Mechanics: children perceive AI as an all-knowing friend, they believe fiction more easily and are more willing to share personal things, and sometimes they stumble upon inappropriate things. The task of an adult is not a ban, but accompaniment and rules. Basic family rules: do not tell AI your last name, school, address, do not send your photos; everything important (lessons, health, “is this true”) should be double-checked with an adult; remember that AI makes mistakes and makes things up - it is not a person and not a friend. Settings: enable child or family mode, a separate account with the child’s age, history under parental control, time limits. A technique instead of total control is the “traffic light rule”: green you can ask yourself, yellow (personal, shopping) - with an adult, red (address, photo, meetings) - never. Second trick: once a week, together analyze one AI answer and look for a possible mistake in it - this is how the child learns criticism, and not blind trust. Insider: a teenager is scared off by surveillance, but is attracted to the role of an “expert” - let him teach the younger ones the rules. A typical mistake: one ban without explanation - the child will simply switch to another device.
Unlock access to submit solutions for instant AI review.