Perhaps Amazon could implement a Google-type of indexing software. Every ebook is stored on an indexed server, and any books that are uploaded are randomly accessed for several paragraphs, which are them submitted to a search of their database. Regardless of the submitted name, author, genre, etc.. the server will come back and determine whether the passages of text are found in any other book. If so, more searches are done with the same book. they all come back the same, the book is denied.
It would take a lot to implement, I'm certain.