Is AI safe for kids' learning? Ten questions to ask any app

Whether AI is safe for a child's learning depends on how the app is built, not on the word "AI". Ten questions sort most apps quickly: can the child chat with it, does it listen, who sees what they make, what is stored, what screens run, who reviews, what leaves the app, whether there are ads, whether you can delete, and whether it ends. Below are those ten questions with Peabody's answers (peabody.kids, free through the end of 2026). No app is perfectly safe, and any claim of that should make you more cautious, not less.

1. Can my child chat with the AI?

An open chat can go anywhere, and children type surprising things. Peabody: only in one place. The idea box is one-shot: ideas go in, a story comes out. The one chat is Talk with Peabody, a short typed chat for 3rd grade and up in a family account: every message is checked before any model sees it, every reply is checked twice, and the family's grown-ups can read it. Kids can send each other phrases from Peabody's own list ("Hi!", a reaction face) by tapping, never typing, and a grown-up can turn that off. The one place kids type to each other is the online co-op in the Doomcrawl game, where every line is screened first (no names, numbers, places, links or rude words get through) and a grown-up chooses who their child can type to.

2. Does it listen?

Microphones in a child's app deserve a hard look. Peabody: nothing is recorded, and Peabody never listens to a child read. The 🔊 buttons use the browser's own voice to read text aloud. A 🎤 button in a few boxes turns speech into typed words: the browser's own speech recognizer is tried first, which in browsers like Chrome and Safari may send the sound to the browser maker to do that, and where it is missing a small speech model (Whisper) runs on the device instead. Peabody never records or receives audio.

3. Who can see what my child makes?

Peabody: a child's stories stay on their own shelf and never go into the public library. A story can be read by its number, like peabody.kids/story/1234, only after it has passed the screens and only if it's gentle or a little spooky. That page shows the pen name only and is kept out of search engines. See sharing a story by number.

4. Does it use my child's real name?

Peabody: no name is needed. One tap makes a reader with a random pen name and face, and there's no email, age, birthday or password. For a child without an account, no name is sent to the AI.

5. What screens run before a child sees a story?

Look for more than one layer. Peabody: an idea screen first, hard limits in the writing instructions, code checks for contact details, links and bad words, and an AI screen that passes, flags or blocks a story and labels how scary it is. A stronger model gives a second opinion on anything unclear. The full list is in is an AI story generator safe for kids?.

6. Who reviews what's flagged?

Peabody: anything still in question is held for a person to review and shown to no one until then. We don't say every story is read by a person, because it isn't.

7. Where does the data go?

Every AI app uses outside services, so ask for the list. Peabody: four outside services for children's data. Anthropic, whose AI model writes and screens the stories (it receives the ideas, level, practice words and mood). Supabase, which stores the data. Vercel, which hosts the site and counts visits without cookies. Resend, which sends email to grown-ups with accounts. A child's ideas are deleted once the story has been screened. The privacy page has the details, though it is still a draft awaiting a lawyer's review.

8. Are there ads or data sales?

Peabody: no ads, and it doesn't sell data.

9. Can I delete it?

Peabody: a reader without a name can delete their Peabody in two taps, and clearing the browser's cookies ends that reader on that device. Readers that go unused are removed automatically.

10. Does it end?

Apps that never stop are designed to keep a child in. Peabody: a story has a last page, and a session has a clear "all done". Peabody has points that are earned and not spent, and no shop, and no night-time pushes.

What the checklist can't tell you

• Whether a filter missed something. All filters have gaps. That's why layers matter, and why a grown-up nearby still helps.
• Whether the company will stay careful. Read the privacy page, and revisit it.
• Whether your child is ready. You know them best.

What to do tonight

1. Pick the app you're thinking about and answer the ten questions.
2. Try it yourself first, for five minutes.
3. Sit with your child for the first session.
4. Ask them what they made, and what they liked.

For the privacy side, see your child's data on Peabody, the free AI story generator page and the free AI learning tool for kids page. The welcome page has the plain-language overview, and the library is open to read.

If the app in question is a general chatbot, ChatGPT for kids: is it safe, and what are the rules by age? goes through OpenAI's own rules and settings.

Learn more elsewhere

Start with age. Common Sense Media advises no AI chatbots for children 5 and under, because young children can't fully tell AI from real people, and limited use beside a grown-up from 6 to 12. Peabody has no open chatbot (its one chat is short and starts at 3rd grade), and its stories are written by AI, so the same care applies: with a child under six, read together.

• Common Sense: AI chatbots and your child, by age: no chatbots at 5 and under, limited use beside a grown-up from 6 to 12, supervised use for homework from 13.
• Common Sense AI risk assessments: independent tests of the chatbots and AI products kids actually use.
• NSPCC: talking to children about AI: eight practical steps from the UK children's charity.
• HealthyChildren.org: how AI chatbots affect kids: the American Academy of Pediatrics' parent article.

For about seventy more, checked and sorted by who they're for, see where to learn about AI.

Questions parents ask

Is Peabody perfectly safe?

No app is, and Peabody doesn't claim it. It uses several layers, and anything in question is held for a person to review.

Does Peabody use my child's data to train AI?

Peabody doesn't sell children's data or use it for marketing. How Anthropic handles data it receives under Peabody's account is a question for Anthropic's own terms, so read those alongside the privacy page.

Does my child need an account?

No. One tap makes a reader on that browser, with no name, email or password.

Is it free?

Yes, for everyone through 31 December 2026, with no sign-up and no card.