See what your PDF says about you, then remove it
- Open PDF Metadata and choose the PDF. It is read at once and the page lists everything the file says about itself.
- Read the report: the document information, the XMP packet and what else is inside, such as attachments, scripts and the names in comments. If the first two disagree, the page points it out.
- To keep the file but change a field, use Edit fields. To take it all out, use Remove metadata, and decide whether attachments and JavaScript go too.
- Press the button. The new file is read again, the page lists what is left, and you can download it.
If you only want the removal, Remove PDF metadata opens on that tab.
What a PDF records about itself
The oldest place is the Document Information dictionary, called Info. It holds a title, an author, a subject, keywords, the program that wrote the document (the Creator), the program that made the PDF (the Producer), a creation date, a modification date and a Trapped flag from the print trade. Programs can add entries of their own, such as a company name.
Newer files also carry an XMP packet: a block of XML that says the same things again, and often more, such as document identifiers and an edit history. A viewer may read either place, so they can disagree when a program updates one and forgets the other.
The report also lists attached files, JavaScript, form fields, bookmarks, the names in comments, and private data: notes a design program keeps for itself, metadata on pages and pictures, page thumbnails, and objects left over from saving on top of an older file.
How much a file gives away depends on how it was made
- A word processor or design program usually writes an author name, often copied from the user account or the program's settings, the program's name and version, and the creation and last change dates.
- A web page printed to PDF usually carries the page title, the name of the browser's PDF library and the dates, and little else.
- A screenshot or photo saved as a PDF often carries only the name of the system or app that saved it.
- A scanner app often names the app and sometimes the device.
- A file saved by adding to the end of the previous version can still hold the earlier title and author as objects nothing points to. They do not show in a viewer's properties, but they are in the file.
These are tendencies, so read the report for your own file.
When it matters, and when it does not
- Job applications. A CV can carry a name from the computer's user account rather than the name you apply under, plus a creation date.
- Legal and business documents. Dates and program names can contradict what you say about how and when a document was made, and an old title left in the file can reveal what an earlier draft was called.
- Anonymous submissions, such as a review, a report or a tip to a journalist or a regulator. The author field and the program are an easy way to link a file to a person.
Metadata is only one trail. The words and style, a name in the text, comments, photos inside the file and the way you send it can identify you too. Removing metadata makes a file cleaner, not anonymous.
It does not matter for a flyer, a menu or a paper that carries your name on purpose. Sometimes a title helps readers, and then Edit fields lets you change the author or the title instead of removing everything.
Read it yourself first
Most viewers show some of this in the document properties: Acrobat Reader under File, then Properties, and Preview on a Mac under Tools, then Show Inspector (menus vary by program and version). That usually covers the main Info entries, not the leftover objects or comment authors that the report on PDF Metadata lists.
What removing takes out, and what it cannot hide
Removing means the data is gone from the file, not hidden: the objects are deleted, including old copies that nothing points to.
| Where it hides | Does Remove metadata take it out? |
|---|---|
| Info entries: title, author, subject, keywords, program, dates | Yes |
| XMP packet and its edit history | Yes |
| Metadata on pages and pictures, private program data, thumbnails | Yes |
| Old copies left by earlier saves | Yes |
| Attached files and JavaScript | Only with the separate switch, which is off by default because forms use scripts to calculate (scripts inside an XFA form are not removed) |
| Names in comments | No. Delete the comments in a PDF viewer |
| Camera and GPS data inside a JPEG photo in the PDF | No. Strip the photo before it goes in |
| Names printed on the pages, company names inside fonts | No. Change the source document |
| The signer of a digital signature | No, and removing metadata breaks the signature anyway |
The file's own name is not inside the file, but it can say as much as any entry. Check it before you send the file.
Photos inside a PDF keep their own data
Many programs store a JPEG photo in a PDF as the original photo file, which can carry its own camera model and the place and time it was taken. That is not PDF metadata, so Remove metadata leaves it, and the report counts the photos that carry such data. Remove it from the photo first, as in the guide to removing location data from photos. JPG to PDF cuts the camera details and location out of each JPG before it stores it (if a damaged file cannot be cleaned safely, it is stored as it is and the result says when a location is still in it), as the guide to saving a picture as a PDF explains.
What changes in the file, and how to check it
The whole file is written again. It is saved as PDF 1.7, the fast web view arrangement is lost (the file still opens normally), and a digital signature stops being valid. A file with a lot of leftover data can come out much smaller. The pages are not touched: the tool compares a fingerprint of every page's content and size before and after, and if a check fails the file is not offered for download. PDFs with a password or permission settings are not supported.
For a check of your own, open the new file in a PDF viewer and look at its document properties. Author, title, creator and producer should be empty and the dates blank. A second program reads the file with different code, so it gives an independent answer.