Is ChatGPT safe for kids?

A consumer chatbot is not a school counselor and not a diary with a lock. Prompts can include names, grades, and family stories. Those strings may be logged by a vendor. 'Safe for kids' is a product and legal question, not a vibe.

Who operates the tool? What is stored? Can a parent export or delete it? Is there learner-to-learner chat? Are outputs filtered by age? WonderMind's answers on this evaluation site: first name and avatar, hashed PIN, no child DMs, stage-aware filters, in-product export on every plan. Legal pages are still drafts — that honesty is part of the safety story.

Sit with a learner and highlight anything in a prompt that would be awkward on a poster in the cafeteria. Rewrite the prompt so the skill remains and the identifiers leave. That is data minimization as a literacy drill, not a terms-of-service shrug.

Crisis talk belongs with a trusted adult, not with a model. If a child discloses harm, WonderMind is designed to drop the lesson and point to a person. A mailbox is not a substitute for a staffed queue — production operators still have to sit in that chair.