If URL "hacking" is illegal, then we have decided as a society that persuading a piece of software to do something does not equate to informed consent on the part of the person operating it (and by extension that we're meant to make some sort of guess as to what they do intend).
Users on the other hand are just people. Being a user on a service does not grant you power over other users (not out-of-the-box anyway). Sure you can scan for vulns and/or follow a public link to a top-secret document: in my view that's nothing wrong. Now reverse the situation: why should a remote server administrator dictate the computing performed on your machine?! Is it ethical for some website operators to start scanning your local network?
A service operator knows about threats, hopefully has counter-measures in place, and can always ban you (or specific requests) if it comes to that. A user is mostly helpless, especially when it comes to computations performed by a script you unconsciously downloaded and executed from a server. How many users are aware of what RCE even means and that a web browser with JS enabled is essentially RCE-as-a-service?
A document that is expected to be executed by the receiving system. Might as well log in to your companies production server as root and send the completely meaningless string of letters "rm -rf / \n", not your fault if the receiving system actually executes it.
You’re not wrong that one could logically evolve your interpretation from circumstances, but GDPR’s authors chose not to accept your interpretation as sufficient, and went further than that.
Now from a practical standpoint, I'd like to see a privacy consent header which informs the web site of the privacy options the user has selected. Absence of that header or absence of specific selections will result in annoying popups like we have now.