I think the way the EU approached this with their "digital gatekeepers" is smart. Recognize that policing the entire internet isn't possible or even desirable. Focus on those few companies with the largest capacity for harm. Different criteria might be appropriate when focusing on potential harm for children (e.g. Roblox rather than Twitter) but besides a few changes you'll probably end up with roughly the same list.
I'm not sure I'd support an outright ban, but rather very strict monitoring and requirements around moderation, in app purchasing, gambling mechanics, and so on.