Also, they seem to fail to recognize that everything subject to copyright is copyrighted on creation, so if they were actually filtering out "copyrighted material", most of the content they are storing would be redacted.