They can do that on Facebook because it’s their code and their platform. They can probably do that on their app and and instance too to some extent but I don’t think they can grab much more than the content of your messages and your likes if you’re on a different instance. Lemmy is open source; if there was a way to get that data we’d know about it.
I’ve hardly managed to move my Instagram posts there and now they’re already rolling in the features I came there to escape.