What your PDF quietly says about you
Every PDF carries a block of text you did not write: which program made it, when, and often who. Including ours.
Every PDF you have ever sent carries a small block of text you did not write and have probably never read. Not the document itself — a separate table inside the file, recording which program produced it, which program it started life in, when it was made, when it was last touched and, very often, your name. What PDF metadata reveals is rarely dramatic. It is just reliably more than people expect, and it travels with the file wherever the file goes.
What is actually in there
The information dictionary holds a handful of named entries, and six of them matter. Title and Author are the two people know about, and Author is frequently your full name or your organisation's account name, filled in automatically by whatever program you used and never mentioned again. Producer names the software that wrote the actual bytes. Creator names the software the document started life in. CreationDate and ModDate are timestamps, accurate to the second, usually with your timezone offset attached to them.
None of that is hidden. Any viewer will show you all of it under document properties, and doing that to your own files once is worth the thirty seconds, because the results tend to be informative. A PDF supposedly drafted last week with a creation date from 2019 is telling you it came off an old template. A tender document whose Author is a rival firm's employee is telling you something rather more interesting.
Producer is a signature, and it is the tool's, not yours
Producer records whichever library last serialised the file, and it is set without anybody asking you. Run a document through almost any tool and the original Producer is replaced by the new one — the string saying "made with Word" becomes a string saying "made with something else". For anyone who cares where a document came from, Producer is the field they read first, and it reports the last tool to touch the file rather than the one that made it.
Here is what this site writes into your document, which I checked by reading the output back with a deliberately non-mutating reader — the ordinary one rewrites these fields as it opens them, which is itself the problem under discussion. Take a PDF whose Producer says "AcmeWriter 9.2" and run it through merge, rotate, add page numbers or extract pages. Every one of them hands back a file whose Producer is now the name and GitHub URL of the underlying library. Not BabaPDF. The library underneath, advertised inside your document, on a site that never mentions it anywhere.
That happens at load, not at save. The library stamps Producer in its constructor, so the original is gone the instant the file is opened — before any tool code runs, whether or not you change a single thing. The modification date goes the same way: I opened a file whose ModDate said 2019 and read it straight back without calling any tool at all. It said today.
Which means there is very nearly no read-only path here. Almost every tool that hands you back a PDF rewrites your Producer and your modification date as a side effect of opening the file, including the ones that are conceptually lossless. A rotate is not a neutral act at the metadata level, however neutral it looks at the page level. Two things escape it: the tools that never build a PDF at all, because turning pages into images hands you images and never writes a document — and encrypted files, which are opened by a route that leaves the existing metadata untouched, so a permission-restricted PDF keeps the producer string it came with.
The archiving irony
PDF to PDF/A is the one tool here that stamps its own name, and it is the tool where that hurts most. The archive function sets Producer to "BabaPDF" and Creator to "BabaPDF", both unconditionally, on every file that goes through it. Your "AcmeWriter 9.2" and your "Acme Desktop Publisher" are both overwritten with a brand name that says nothing whatsoever about where the document came from.
People convert files for archiving precisely because provenance matters to them. Archiving is the act of saying this document should still mean something in ten years, and the Creator string — the name of the program somebody actually drafted it in — is a genuine part of that meaning. Converting a PDF for archiving on this site destroys it, and replaces it with the name of the tool that did the destroying. If the original Producer or Creator matters to you, write both down before you run PDF to PDF/A, because you are not getting them back afterwards.
The same function is careful in the other direction, which makes the contrast sharper rather than softer. It sets a creation date only when the document has none: I ran a file created in March 2019 through it and the creation date came out as March 2019, untouched. Somebody sat down and thought carefully about preserving history in one field, then overwrote it in two others.
What survives which tool
Title and Author survive some tools and not others, and the split is not intuitive from the outside. Rotate, add page numbers and PDF/A all keep them. Merge, split, extract, delete and organize lose them completely — title and author come back empty, because those tools build a new document and copy pages into it, and the title was never on a page. The same mechanism empties your form fields, and it is worth reading if you have ever wondered where your bookmarks went.
Creator splits differently again. Rotate and page numbers keep the original Creator, while merge and extract replace it with the library's name. The rule underneath is that the library only fills Creator in when it finds the field empty — and a freshly rebuilt document has an empty one.
- Rotate, add page numbers: title, author and creator kept. Producer replaced. Modification date replaced.
- Merge, split, extract, delete, organize: title and author gone entirely. Producer and creator replaced with the library that did the rebuilding. Creation date quietly reset to today.
- PDF to PDF/A: title and author kept. Producer and creator both overwritten with "BabaPDF". Creation date preserved, if there was one to preserve.
- Protecting a PDF: everything kept. Title and author are encrypted along with the pages, so anyone holding the password reads them exactly as they were.
- Removing a password: everything kept. This one is worth a footnote — it used to lose the title and author silently, which is exactly the kind of thing you would never notice until the day you needed the field.
- Flattening to images: title, author and creation date gone. The result is a new document built out of photographs of the old one, so nothing came across with the pixels — and it is stamped with the library's producer string and today's date like any other new file. The other flatten mode, the one that only locks form fields, mutates in place and keeps everything.
- Any encrypted input, whatever the tool: title and author kept — and the producer string kept too, which is the one place this site does not overwrite it. Encrypted files are opened by a different route that leaves the existing metadata alone, so a permission-restricted PDF comes out of a rotate still naming the scanner that made it.
There is no metadata editor here, and that is a gap
BabaPDF has twenty-four tools and not one of them lets you read or edit metadata. You cannot see your own Author field, you cannot clear it, and you cannot set a title. That is a real gap, and it is better said plainly than left for you to discover: if you need your name out of a document before you send it, this site will not do that for you today. Any desktop PDF viewer will, under document properties, in about four clicks.
What can be said for the tools here is narrow but true: none of that metadata is transmitted anywhere, because nothing is transmitted anywhere. Your Author field gets rewritten in your own browser or not at all. That is not the same thing as being removed, and it should not be mistaken for it.
Before you send a file
- Open document properties on your own PDFs once and look at Author. That is what every recipient can see.
- Assume Producer names the last tool that touched the file, not the one that made it. It is evidence of handling, not of origin.
- Timestamps are precise and frequently carry a timezone. A creation date can contradict a story on its own.
- If provenance is the reason you are archiving, record the original Producer and Creator first — PDF/A conversion overwrites both of them.
- Merging or splitting resets your creation date to today and empties the title. That is not a privacy win. It is just wrong.
