There’s a discussion over in the EFS-Web listserv (the listserv for patent practitioners) about the best way to create a PDF so that it won’t unnecessarily reveal metadata. The answer for Windows users is of course to use CutePDF to create the PDF. Which then led to the question of the extent to which USPTO strips out metadata in which case maybe the filer can be lax about this. What is the answer?
Part of the answer comes from the fact that the EFS-Web system converts your uploaded PDF into a very poor quality (half-toned) image PDF. This is what is made visible in IFW. Because it is image-based, your various metadata will be lost.
But there are two trouble areas for the metadata in your uploaded PDF.
The first is that if you indexed it as a non-black-and-white drawing (in a utility case) or as a drawing (in a design case) then it will find its way into the SCORE system and will be visible under the Supplemental Content tab. The SCORE files are complete and exact copies of what was filed. So any metadata in such a file will be visible in the SCORE file and will be available to any member of the public who has the energy to go hunting in SCORE.
The second is that the USPTO preserves every file that you uploaded, complete with metadata, in a place that you can’t see in PAIR. So even if a member of the general public cannot get their hands on the original PDF (the one that has all of the metadata still inside), people at the USPTO certainly can get their hands on it.