This is not really a specific item, but more curiosity from recent security articles. Basically, they have proven the ability to intercept signals via cross talk between USB ports as well as from the computer itself. I remember when I got my last interface I was paranoid to use a USB port for this reason, so went with fire wire to get some semblance of isolation.
 
Has anyone seen this studied specifically relating to audio interfaces? It makes me wonder sometimes if system setup with high USB count (not hubs) plays into some problems that folks see. I have never specifically seen any information on tolerances and noise removal (within drivers), but it is curious as to how much cross talk is really going on that is visible to an unused port.

With a quality USB audio interface, I wouldn't give it a second thought.
 
Some newer motherboards have USB ports specifically for use with A/D D/A converters.
Cleaner power... 
 
I measured the average noise-floor with a RME Fireface UFX connected to both normal USB ports... and the A/D specific "clean power" USB ports.  Absolutely *zero* difference in the average noise-floor.  Zero difference in performance.  
 

Thanks Jim. I will need to research more when my computer finally bellies up. I didn't realize that newer motherboards had dedicated USB specifically for A/D D/A.
 
I should have stuck this in the OP but most of the articles stem from this study which was published on August 10th. The specifics of testing is not given, but the "90% of computers and hubs tested (45 out of 50) had [exploitable] USB leakage." That number stuck out like a sore thumb, so I was curious how manufacturers "deal with it." The details of the testing would certainly help to understand the results, and it obviously has more application to security than to audio, per se.