^ Couldn't we also say the that noise is not actively filtered out but - instead - gets filtered out? This is because in the digital world neither the sender nor the receiver can tell the difference between a noisy 1 (or 0) and a clean one.
The receiver that extracts the digital data doesn't care (up to a point). Question is what happens to that noise with respect to the rest of the DAC. If allowed to bleed into the DAC reference, clock or analog output, it would show up there.
Fortunately vast majority of DACs -- even the ultra cheap ones -- are immune to this to great extent. And certainly below threshold of audibility.
There are exceptions such as the horribly designed Schiit Modi where if you cough next door, its output changes.
Using that and Sonore MicroRendu I did a bunch of
measurements of USB cables: Here is an example outcome, comparing the TotalDAC d1 "audiophile" USB cable to generic but long one:
But if you make the generic cable short (yellow), it handily beats the TotalDAC cable (red):
Switching the DAC to Behringer we see that even the long cable does the same thing as TotalDAC:
I did bunch more
tests of USB cables including "stub" rigid connectors from likes of Uptone:
Here we see that with sensitive DACs, shorter USB cables is better. But again with half decent DACs, that doesn't matter either:
I let you read the article for the rest. Bottom line, if you worry about USB cables, just keep their lengths short. Better yet, get a DAC that is insensitive to such things which per above, includes vast majority of DACs.