SafetyFamilies

How are AI answers checked?

Inputs and outputs pass a stage-aware filter. Younger stages buffer the full reply before it appears on screen.

Updated 8/23/2026

Two layers

Provider moderation plus WonderMind's own stage-aware rules. Sprout and Explorer replies are screened before display. Older stages stream, then log problems after the fact.

The Guide can still be wrong

Fluency is not a fact check. For dates, medical advice, legal claims, and school answers that will be submitted, verify with a human source. The product teaches that habit on purpose.

Blocked subjects

Parents can hide topics. Hidden subjects never reach the model. If something still feels off, use the safety mailbox rather than arguing with the bot.

How do parent controls work?
From a child's report you can set time limits, hidden subjects, exploration access, sensitive-topic alerts, and Co-pilot mode.
Why won't WonderMind do the homework?
The product teaches AI literacy. It will help you outline, critique, and research — not generate work to submit as your own.
What happens if a learner is in crisis?
WonderMind answers with care, drops the lesson, and points to a trusted adult or professional help. It is not a therapist.

All help articles Email support

How are AI answers checked?

Related