Problem1:

copyrighted content is not identifyable to cache processes yet. Content could be anything from pictures, texts, software, audio data - you name it.

Problem2:

statistics of hits on original source incorrect (example: a Web-publishing company gets their fees based on page-accesses/popularity)