1. Can you see what your child sees?
You should be able to open any of your child's conversations. A tool that hides history from parents, or lets it be deleted silently, fails the first test.
2. Will you be told when something is wrong?
Kids sometimes bring big things to a chatbot before they bring them to a person: bullying, distress, self-harm, inappropriate content. A child-safe tool should flag those conversations to the parent, not just quietly refuse.
In Cornerstone, every member message is screened, and flagged conversations appear in the parent's alerts panel.
3. Does it admit when it doesn't know?
AI can state wrong things confidently. For learning, that's worse than no answer. Prefer a tutor that shows its sources and says plainly when a question can't be answered well.
4. Who owns the data?
Check whether you can export your family's conversations and delete the account completely. 'Delete' should mean the data is gone, not hidden.
- Export of all conversations in a readable format
- Full account deletion, not just deactivation
- No child email or password required
5. Does each child have their own space?
Siblings sharing one login means mixed histories and no per-child oversight. Separate member profiles keep each child's learning — and each child's alerts — distinct.
Quick answers
Are general AI chatbots designed for children?
Generally no — most are built for adult users and lack parent visibility and safety alerts.
What is the single most important safety feature?
Parent visibility: being able to see every conversation, combined with alerts when one needs attention.