The Freenet has a wild array of content and much of it is not particularly interesting to the common man. And the web interface is kind of not so user-friendly. But the mechanism itself is great: you could use Freenet as an anonymous shared data store for your application.
For example, PirateBay could move to live on Freenet and have its own client (either a local application or a locally run webapp) just fetch the torrent listings from Freenet data blocks. For now, Freenet can't do big files so that must still be done on the opennet but much of the metadata could be irreversibly stored in the decentralized system that is Freenet.
As computers and the internet get faster and faster, we might be able to sacrifice some speed for safety, anonymity and plausible deniability. At the simplest, big files could be distributed in the open by simply xoring them together with other big files. You'd have to download two big files to construct the one you want. On the other hand, a single file would host information from several files so downloading it couldn't possibly be considered infringing anything. It could just be genuinely random data that, when xored with another file, turns out to be an mp4 of a film.