Five of ten AI assistants have parental controls. Five have none

There is a comfortable assumption: if an app admits a 13 year old, somebody must have built in brakes. In half of the cases that is not true.

The five that do

ChatGPT, Gemini, Copilot, Meta AI and My AI offer some form of adult control. They are not equivalent to each other, but at least they exist.

The five that do not

Claude and Character.AI offer none, and there is a reason: both require 18 anyway. The other three are the real problem. Perplexity, Grok and Canva admit 13 year olds and ship no parental control whatsoever. A 13 year old signs in and there is nothing an adult can adjust from outside.

What ChatGPT actually allows

The adult invites the teenager from the parental control settings, by phone number or email, and the teenager accepts from their own account. From there you can set quiet hours, reduce sensitive content, decide whether memories carry across chats, and switch off image generation. There are also alerts to the adult in narrowly defined situations involving safety.

What no control gives you

Linking accounts does not turn a parent into a reader. The chats are not visible and activity is not tracked live. That surprises many parents, but read it the other way round: it is precisely why a teenager agrees to link at all. A control that showed everything would be refused, and would end in a second account nobody sees.

For Gemini the control lives in Family Link

In Google's world the control is not inside the app but in Family Link: from there you can switch off access to the Gemini apps for children under the applicable age. And a little known detail: an account managed by Family Link cannot reach the Gemini web app.