Before you send it
What hidden information does a PDF carry?
More than the page shows, and usually more than the Properties panel shows. A typical PDF records who made it, with what, and when — and it often records that twice, in two places that do not always agree. Clearing the one you can see is how people end up sending a file that looks anonymous and is not.
1. What a PDF commonly records about itself
| Field | What it tends to contain |
|---|---|
| Author | The name registered to the computer or the Office licence — often a full name, sometimes a work login |
| Creator | The application the document was written in: Word, InDesign, a scanner's software |
| Producer | The software that wrote the PDF, often with an exact version number |
| Created / Modified | Timestamps, usually to the second and with a time zone — which says roughly where the author was |
| Title, Subject, Keywords | Frequently left over from a template, and so naming a different client, project or company |
An export from Word names the machine's registered owner. An export from a desktop-publishing package can name the operator and the workstation. A scan names the model of the scanner. None of it appears on any page.
2. The second copy that "remove metadata" misses
PDF metadata is stored in two places, and clearing the visible one leaves the other naming the author.
- The document information dictionary
- The fields above, in the form a viewer's Properties or Document info panel displays. This is what most "remove metadata" features clear.
- The XMP packet
- A block of XML stored separately inside the file — on the document, and sometimes on individual pages and images too. Viewers do not show it. Adobe applications write the author into both places and read it back from XMP in preference.
So a file can arrive with an empty Properties panel and an XMP packet that still says who made it, when, and with which copy of which application. It looks clean. It is not.
3. Deleted is not always gone
A PDF is allowed to contain objects that nothing points to. Many editing tools "delete" a metadata entry by removing the reference to it and leaving the entry itself in the file. Nothing displays it any more, but open the PDF in a plain text editor and the name is still there.
The test of a metadata tool is therefore not whether the fields look empty afterwards. It is whether the values are absent from the output — which is checkable, by searching the saved file for the name you removed.
4. Hidden information that is not metadata
Clearing metadata perfectly still leaves these, because they are part of the document's content rather than its description:
- Comments and annotations — carry the reviewer's name and a timestamp on each one.
- Form field values — including fields that have been hidden or made read-only.
- Attachments — whole files embedded inside the PDF.
- Text under a black box — a rectangle drawn over text hides it from view and nothing else; the words are still selectable and copyable.
- Photo location — images taken on a phone can carry EXIF data, including GPS coordinates, into the PDF they are placed in.
- Digital signatures — name the signer by design.
Checking a file without sending it anywhere
There is an obvious irony in uploading a sensitive document to a website to find out what it gives away. Filozy's Edit PDF metadata tool reads the file inside your browser, reports what both copies contain, and lists the fields XMP still answers after Properties has been cleared. When it removes something it deletes the object as well as the reference, then re-opens the result and searches it to prove the value has gone. It names annotations, form values, attachments and signatures in its report, but leaves them alone — they are content, not metadata.
For text hidden under boxes, use Redact PDF, which removes what is under a mark rather than covering it. For the step-by-step, see removing metadata before publishing. And to confirm none of this involved an upload, check it yourself.
Common questions
- What hidden information is stored in a PDF file?
- A PDF can carry the author's name, the organisation, the software and version that produced it, creation and modification dates, and sometimes a title or keywords left over from a template. It records these in two places: the document information dictionary, which is what a viewer's Properties panel shows, and an XMP metadata packet, which viewers do not show at all. Beyond metadata, a PDF can hold comments, form values, attachments, text hidden under images or black boxes, and GPS coordinates inside embedded photos.
- If the Properties panel is blank, is the PDF clean?
- Not necessarily. Metadata is stored twice — in the document information dictionary that Properties displays, and in an XMP packet that it does not. Adobe applications write the author into both and read XMP first. A tool that clears only the first leaves a file whose Properties panel is empty while the XMP still names the author, the dates and the application that made it.
- Can deleted metadata still be recovered from a PDF?
- Often, yes. PDF files are allowed to keep objects that nothing refers to any more, and many editing libraries remove the reference to a metadata entry without removing the entry itself. The name is no longer shown anywhere, but it is still in the file and readable in a plain text editor. A thorough tool deletes the object as well as the reference and then searches the output to confirm it has gone.
- Does a scanned PDF contain metadata?
- Usually. Scanners and scanning apps commonly record their make and model as the producing application, along with the scan date. If the scan was made with a phone, the page images may also carry EXIF data from the camera, which can include the location where the photo was taken.
- How do I see all the metadata in a PDF?
- Your viewer's Properties panel shows only the document information dictionary. To see the XMP packet as well you need a tool that reads both and shows where they disagree. Filozy's Edit PDF metadata tool does this in the browser without uploading the file, and it lists fields that XMP still answers after Properties has been cleared.