We have to separate child protection from Internet control so that the "protect the kids" narrative loses its potency. So here's a counter-narrative: we can implement digital child protection without Internet-wide access control, and it requires just 3 simple features that can be implemented in less than a week. There's no need to introduce new laws at all. This could just be done tomorrow if there is genuine will to…
So basically parents set the child's age and apps rely on that if they need to know if the user is old enough? That's pretty similar to the California bill. Parents set an age when creating a child's account. The OS provides an API to get the user's age bracket from that, which apps that need to know the age bracket of the user can call.
edit: on second thought, realistically, the API solution is too brittle regardless of which way it goes. Because the API requires every service to implement it and that's not happening, whereas an app installation lock only requires one child-friendly OS to implement it, then parents can choose that OS.