A much better approach would be to hold platforms responsible if they allow a stranger that does not have explicit parental consent to communicate with or get information about a minor.

This would block the most common classes of abuse on platforms like Roblox, Fortnight, Lego (kids) Fortnight, YouTube Kids, Minecraft, and "educational" social networks / games.

Note that it doesn't require any centralized surveillance at all. Parents just need to control the kids' ability to create random accounts, by (for example) turning on parental controls as they already exist on most tablets/phones, and blocking app installation / email applications (or other 2FA vectors).

When the parent allows an account to be created, they just tick the "kid mode" box. This even works with shared devices that don't support multiple accounts (so, iPads and iPhones).