The PDF file format is very versatile, and one of the things it can contain is accessibility information. (Metadata for the benefit of screen readers, etc.)
On our Sitecore 8 install, we have PDFs that got uploaded via Content Editor and published to the site, but when subsequently downloaded from the site are missing their accessibility metadata. We've verified that the accessibility data was there before the upload; the file that's coming out is not the same as the file that went in.
I'm fairly new to Sitecore work, but I'm guessing that this is the result of an overzealous filter of some sort. PDFs are versatile enough to embed malware into, so it would be reasonable to want to run uploaded PDFs through some sort of white-list to strip out anything not known to be legitimate, but I could easily be wrong about what's happening here.
Either way, how can I find out what's going on and fix it?
Googling sitecore pdf removing accessibility data and several variations on that basic theme results in nothing useful. Hopefully someone here will know how to deal with this...
