Mumbot Resources The Reset
Guide · Secure AI

Four rules for kids and AI

The settings that make them stick, and the exact paths to find them.

One: the account stays in your name

Whoever owns the login controls the settings, so set it up on yours rather than theirs. It means you can check what's switched on whenever you like, without having to ask for their phone.

Two: decide about training

You'll find it under Settings › Privacy › Help improve Claude. Leave it off and nothing new you write goes into training the next model. Switch it on and those conversations can sit in the training pipeline for up to five years.

Worth clearing up, because the internet gets this wrong constantly: your chat history stays either way. It only clears when you delete it. The thirty-day figure people quote is about how long deleted chats take to clear the back end, not an expiry date on your history.

Three: name what it actually is

Somewhere to think out loud and get unstuck, rather than an oracle handing down the right answer. Kids who understand that difference ask better questions and take the output far less literally.

Four: it can be confidently wrong

The tricky part is that it sounds just as certain when it's wrong as when it's right. Teaching them to check anything that matters does more for them than any toggle on this list.

One more thing

Rating a reply with thumbs up or thumbs down sends that conversation to Anthropic as feedback, and feedback is kept for five years. Worth knowing before anyone starts tapping the little icons.