[derefr]

How are these not also problems with the original service, though? If a judge would say that "they know what they downloaded", then why wouldn't they "know what they're proxying"? Since both processes would be automated, knowledge of either would require explicit logging. (And before you say that the downloading step enables a point where the entire illegal product is, for however brief an instant, on-disk where logging could be much more easily applied, consider this alternative setup: the company just rents out "CPU time" to their customers to run virtual machines running with encrypted disk images, a known image of which happens, as its effect, to download a torrent, encrypt it, and dump the result to a write-only host-shared folder. They then send their customers any such "dumped logs" on disks as a courtesy service.)