The Parents' Guide to Kids and AI Safety

About this page: A year of writing about children and AI has passed thirty articles, which created a new problem: where do I start? This is the map. Each topic gets its conclusion in three lines, with the full article linked for when you need the depth. Built to be bookmarked and entered wherever your question is.

1. Is it safe at all?

In three lines: “AI” can’t be judged safe or unsafe as a lump. Nearly all the real risk collapses into one situation: an adult-oriented AI used by a child alone. A tool designed for children, used where a parent can see, keeps the risks in manageable territory.

• The full picture → Is AI Safe for Kids?
• Our own safety design → Our Safety Commitment

2. From what age?

In three lines: General chatbots (ChatGPT and peers) require 13+ by their terms; below that, “on a parent’s account, together” is the only sound form. Purpose-built kids’ voice AI works from roughly the age a child can ask a question aloud — about 3–4.

• The age-by-age map → What Age Can Kids Use AI?
• ChatGPT specifically → What Age Is ChatGPT OK For?

3. How do we choose?

In three lines: “AI apps for kids” is really five different product types. Sort the type first, then filter with six questions: built for children? voice data kept out of training? history readable by parents? ad-free? usage-capped? a real company behind it?

• The five types, compared honestly → Safe AI Apps for Kids
• The checklist → How to Choose a Safe AI App
• Versus the smart speaker you own → AI Speaker for Kids vs. Alexa
• If the goal is English practice → Choosing an English App
• Sharing it as a family → Family AI Apps: the four conditions
• Chat-shaped AI specifically → AI Chat for Kids

4. Where does my child’s data go?

In three lines: Three questions settle it: is the voice used to train models? can parents read the conversations? does deletion actually delete? For us: ten-second recordings, never stored, processed in Japan, never used for training. A service that can’t answer those three has answered them.

• The full journey of one recording → Where Does My Child’s Voice Go?
• The formal version → Privacy Policy

5. Doesn’t AI get things wrong?

In three lines: Yes — in a confident voice, which is the dangerous part. But most of what young children ask sits in low-risk territory, and the real protection isn’t a filter: it’s the habit of “shall we check that?” — learned at seven, useful for life.

• The mechanics and the household habit → What If the AI Tells My Child Something Wrong?
• The homework boundary → Summer Projects with AI

6. Worried about screens and dependence

In three lines: Watch whether the screen waits for your child, not the total minutes. And check any kids’ app for a usage cap, and for the absence of streaks and “don’t go!” mechanics — designs that maximize time-on-app and designs that serve children do not coexist.

• Passive vs. active → YouTube vs. Talking
• How to count the time → How Much Screen Time Is OK?
• What even counts → What Counts as Screen Time?
• The “my AI friend” question → “Poppy Is My Friend” — Is That Okay?

7. Conversation, language, and growing up

In three lines: What builds a child’s language is conversational back-and-forth, not word volume. AI can generate turns structurally — and is not a substitute for talking with people, a sentence we apply to our own product first. An addition, never a replacement.

• The research → It’s the Turns, Not the Words
• Why voice matters pre-reading → Voice-First AI for Kids
• The unanswerable questions → Handling the Questions You Can’t Answer · “Why Do I Have to Go to School?”
• What children actually ask → Three Findings From the Logs

About this guide’s vantage point

This guide is written by the makers of KidTalk, a kids’ voice-AI app — so of course we have a position. That’s exactly why every article states what our product cannot do: AI makes mistakes, it doesn’t replace human conversation, it doesn’t replace professional support. Keeping everything checkable before asking for trust is our version of honesty.

This page is updated as new articles land. Questions → contact us.