Why everyone believes it
The reasoning is real, it's just being used to answer the wrong question. When your child sends a message to ChatGPT, or any AI chatbot, that message is encrypted while it travels from their device to the AI's servers and back. Nobody sitting in the middle of that journey, not your broadband provider, not anyone on the same wifi, not a stranger on the network, can intercept it and read it on the way.
That part's genuinely true, and it's a good thing. It's the same encryption protecting your online banking.
But it's not the only way to see a conversation.
What that explanation leaves out
Encryption only protects a message while it's travelling. The moment it arrives and shows up on the screen, in the chat window, it has to turn back into ordinary, readable text. That's the whole point of it getting there, your child couldn't read a reply from ChatGPT that was still scrambled.
Once it's sitting there on screen in plain text, anything with permission to look at that screen can read it too. Nobody's cracking any encryption. Nobody's intercepting anything mid-flight. It's closer to looking over someone's shoulder than hacking into anything, just done by software instead of a person standing behind them.
That's what browser extensions do. One that's been given permission can read the text already rendered on a webpage, the same page your child is looking at. If that page happens to be an AI chat, and the conversation is sitting there in plain text, it can read it.
Why so many parenting guides get this wrong
Search around and you'll find plenty of advice sites confidently saying no tool can do this, that it's technically impossible. A lot of that advice hasn't caught up. It's reasoning correctly about encrypted connections and applying it to a completely different problem, what's visible on a rendered screen.
It's a bit like saying nobody can read a letter once it's delivered, because the postal service can't open sealed envelopes in transit. True of the envelope. Not true once it's opened and sitting on the kitchen table.
Does this work on every device?
Not equally. It's easiest on a computer, through a browser, because that's where extensions can be given permission to read a page. On a phone, inside a dedicated app rather than a browser, it's harder. Most tools that try to monitor AI chats on phones can only see what your child types, not what the AI replies, because they're watching the keyboard rather than reading a rendered page. Seeing both sides of a phone-app conversation the way you can in a browser generally isn't possible yet.
Is this secretive?
Not necessarily, and it shouldn't be. A browser extension has to be installed on the device, and most reputable ones show up as an icon in the toolbar, visible to anyone using that browser, including your child. Covert tools built to hide themselves do exist, but reading a rendered webpage doesn't require secrecy. That's a choice a tool makes, not a technical necessity.
What this actually means for you
If your child is chatting to an AI on a computer through a browser, and something with permission to read that page is installed, it's technically straightforward for it to see the conversation. Nothing's being hacked, nothing illegal is happening, and no encryption is being broken. It's just reading what's already there on screen.
If you've been told it simply can't be done, that answer is out of date.
