It is not that easy. They need a reliable system to do that. For what can go wrong, see all the hate Youtube's ContentID gets. But, identifying whole books should be simple enough.
If the official book is available on Kindle, then it actually should be quite simple for them to use a locality-sensitive hash to find plagiarized uploads.