What PDF Metadata Reveals – and How to Remove It Before You Share

Every PDF carries information you don’t see on the page: who or what created it, which software made it, and when. Usually that’s harmless. But it can reveal a name, an internal department code, or the fact that a “new” document was really made three years ago. Here’s what we found in real PDFs, how to check your own, and what removing metadata does and doesn’t do.

What PDF metadata looks like: a real example

We opened the official IRS Form W-9 in PDFroo’s PDF Metadata Editor:

The Author field says SE:W:CAR:MP, an internal office code rather than a person. Creator and Producer both say Designer 6.5, the authoring software. The creation date is March 6, 2024. IRS Publication 15 lists its author as W:CAR:MP:FP and names two tools by version: Antenna House XSL Formatter and iText 2.1.7. On a file you create yourself, the Author field is often your full name, taken from your Word or Acrobat settings.

Where metadata lives (there are two copies)

A PDF can store metadata in two places:

• The document information dictionary: Title, Author, Subject, Keywords, Creator, Producer, and creation and modification dates. This is what most viewers show.
• An XMP metadata stream: an XML block based on Adobe’s Extensible Metadata Platform, now the ISO 16684-1 standard (Adobe XMP documentation). The W-9’s XMP block is about 4 KB. It repeats the title, description and dates and adds unique document and instance IDs.

That duplication matters. In our testing, we found that editing only the visible fields can leave the old values in the XMP copy. That was a bug in our own editor: it updated the visible title, but the XMP still said “Form W-9 (Rev. March 2024)”. We fixed it on October 8, 2026, and PDFroo now removes the stale XMP block whenever you save edited metadata.

How to check a PDF’s metadata

• Google Chrome: open the PDF in Chrome, click the ⋮ menu in the PDF toolbar, then choose Document properties.
• Adobe Acrobat Reader: File → Properties (Ctrl+D on Windows, ⌘D on Mac).
• Preview on Mac: Tools → Show Inspector (⌘I).
• Any browser: drop the file into the PDF Metadata Editor. It shows every field plus the dates and page count, and the file stays on your device.

How to remove PDF metadata

1. Open the PDF Metadata Editor and add your PDF.
2. Tick Remove all metadata, or edit individual fields instead, for example to replace your name with your company’s.
3. Click Save PDF and download the cleaned copy.

We verified the result with Poppler’s pdfinfo. Before, the W-9 reported a title, subject, keywords, author, creator, producer, both dates, and “Metadata Stream: yes”. After “Remove all metadata”, it reported none of those fields and “Metadata Stream: no”. All 23 form fields were still present, and nothing on the pages changed. Publication 15 gave the same result.

Before                              After
Title:    Form W-9 (Rev. March 2024)  (none)
Author:   SE:W:CAR:MP                 (none)
Creator:  Designer 6.5                (none)
Producer: Designer 6.5                (none)
Metadata Stream: yes                Metadata Stream: no

What removing metadata does not remove

Metadata is only one kind of hidden information. In our tests, these stayed after stripping metadata, as expected:

• The file identifier. PDFs carry an /ID value in the file trailer. It isn’t personal, but it stays.
• Everything on the page, including filled-in form data.
• Scripts and embedded data. The W-9 still contained its XFA form data and JavaScript after stripping. Attachments and comments, if a file has them, are separate from metadata too.

Covering or cropping text is not redaction

We tested two common shortcuts, and neither removes text from the file:

• White-out boxes. We drew a white box over “Form W-9 (Rev. March 2024)” with Add Text to PDF. The text vanished visually but could still be copied out of the saved file.
• Cropping. We cropped Publication 15 with Crop PDF. Cropping changes the visible page area only. When we reset the page boundaries of the cropped file, the hidden text (the “Contents” list, for example) came straight back.

To make text truly unrecoverable, use a dedicated redaction tool, or flatten the pages to images with Flatten PDF’s image mode after covering the text. Image mode removes the text layer entirely, so the result is no longer searchable or selectable. Check the output before you share it.

Photos carry metadata too

JPG photos often contain EXIF data such as camera model, date and sometimes GPS location. During our testing, we found that PDFroo’s JPG to PDF embedded the original photo bytes, including a test photo’s GPS coordinates (40°41′21″N 74°02′40″W). We fixed that on October 8, 2026. JPG to PDF now strips EXIF and XMP from photos before embedding them, and our re-test found zero EXIF entries. Resize Image also outputs photos without EXIF.

Before-you-share checklist

• Check Document properties for your name, company or old dates.
• Use Remove all metadata, or edit Author to what you want shown.
• Remove pages you don’t need with Delete PDF Pages.
• Hide sensitive text with real redaction or image flattening, not white boxes or cropping.
• For sensitive files you email, add a password with Protect PDF and send the password separately.

Want to know what happens to files in online tools generally? Read Are online PDF tools safe?

How we tested

All tests ran on October 8, 2026, using PDFroo’s production code in Google Chrome on a desktop computer, with public IRS files and test photos we created ourselves (the GPS coordinates are fake). Outputs were inspected with Poppler (pdfinfo) and PyMuPDF. The white-out and crop results come from extracting text from the saved files.