With ubiquitous surveillance, you can build a database of the location activity of all citizens. This is problematic for several reasons, with top-of-mind right now being targeting individuals for political reasons, but also stuff like what if the database gets hacked, and general fourth- and fifth-amendment concerns.

With a ban in place, the ability to build such a database drops significantly because you need a pretense for every image or piece of video you acquire. You can acquire and send for processing security camera footage when there was a crime, but when there was no crime? How do you justify acquiring the security footage, and how do you justify asking for it to be processed. Whether the footage available to facial recognize is ubiquitous or not makes a qualitative impact.