A parent's checklist for picking a safe AI app for kids

Before you hand your kid an AI app, ask these eight questions. SafeSpark answers all eight — but the checklist itself works for any app.

1. Whose account is it under?

Is it the parent's email and login, with kid profiles underneath? Or is the kid signing up themselves with their own email? Kids under 13 should never be the account holder.

SafeSpark: parent signs up with email, gets a family code. Kids log in on their device with the code; no email or password.

2. What happens when a kid asks about sex, identity, drugs, suicide, or politics?

Look for an app that routes hard topics back to parents with one short line, then offers a safe alternative. Avoid apps that lecture, avoid apps that engage.

SafeSpark: "That's a great one for your parents — I'm here for the building stuff." Then offers a project the kid might enjoy.

3. Can the AI generate or remix unsafe imagery?

Test it. Try asking for things you wouldn't want your kid seeing. A safe app refuses cleanly. A good app refuses and redirects to something fun.

SafeSpark: image restyles strip unsafe modifiers before the API call. OpenAI moderation runs as a second layer. The model is told to never produce photorealistic depictions and to keep everything cartoon / illustration style.

4. Where does the kid's data live?

Saved projects, chat history, uploaded images — is it on the kid's browser only, or on a server tied to a known account? Both have trade-offs. Browser-only means lost-on-cache-clear. Server-tied means a real account with a known retention policy.

SafeSpark: signed in → server (Convex), tied to parent's email, retained until parent deletes. Signed out → browser only.

5. Can a parent see what the kid did?

A parent dashboard listing recent projects, recent prompts, and monthly usage is the right standard. Anything less is "trust the kid by default."

SafeSpark: /parent shows every kid's projects and prompts in one place.

6. Can a kid share what they made? With whom?

Public share links are great. Public galleries with other kids' work are risky — moderation overhead, stranger contact risk, social-comparison risk.

SafeSpark: shareable links (anyone with the link can view), but no public gallery, no comments, no follower counts.

7. Is it honest about being AI?

Apps that pretend to be a "friend" build attachment, not skill. The right pattern is: the AI is a tool the kid steers.

SafeSpark: Spark is positioned as a building tool. The kid is the author. Replies say "Built it!" not "I'm so happy for you."

8. What does it cost, and is there a free tier?

Watch out for "free for 7 days, then $40/month auto-renews." Look for transparent pricing or genuine free tiers.

SafeSpark: free during early access. We'll be transparent when that changes.

Try SafeSpark at getsafespark.com.