How to remove PDF and image metadata before you send a file
Redacting the page and forgetting the metadata is the second most common way a document gives away what it was meant to hide. The page is what you looked at. The metadata is what you never opened.
What a PDF is carrying
- The document information dictionary — Title, Author, Subject, Keywords, Creator, Producer, CreationDate, ModDate. Author is very often a real person's account name; Title is frequently the original filename, including the bit you renamed the file to hide.
- The XMP packet — a second, independent copy of much of the same thing, plus editing history in some tools. Clearing the info dictionary and leaving XMP is a common half-fix.
- Embedded originals — attached files, embedded fonts with subset names, and images that still hold their own EXIF.
- Previous revisions — PDFs can be saved incrementally, keeping the earlier version of the document inside the same file.
What a photo or screenshot is carrying
- GPS coordinates — latitude, longitude and often altitude, written by nearly every phone camera unless location access was denied.
- Device and lens — make, model, serial number in some bodies, lens, and the exact capture settings.
- Timestamps — original, digitised and modified, usually in local time, which also discloses the time zone.
- Thumbnails — EXIF can embed a small preview that was generated before your edits. Cropping a face out of a photo does not always crop it out of the embedded thumbnail.
- Software trail — what edited it and when.
Read it first, so you know what you are removing
exiftool file.pdf # or file.jpg — prints every tag pdfinfo file.pdf # PDF info dictionary only strings file.pdf | head -50
If you have no tools installed: on Windows, right-click the file, Properties, Details tab, and use Remove Properties and Personal Information. On macOS, Get Info shows only a fraction of it, so do not trust a clean-looking Get Info panel.
Removing it
| Route | Command or step | Note |
|---|---|---|
| exiftool | exiftool -all= file.jpg | Thorough for images. Keeps a backup as file.jpg_original unless you add -overwrite_original. |
| exiftool, PDFs | exiftool -all= -overwrite_original file.pdf | PDF is append-only by design, so exiftool marks the old metadata as deleted rather than truly removing it. Use qpdf --linearize afterwards to write a clean file. |
| qpdf | qpdf --linearize in.pdf out.pdf | Rewrites the file structure, dropping unreferenced objects and earlier revisions. |
| Re-render the pages | Rasterise and rebuild | The most complete option: nothing from the original file structure survives into the new one. |
| Windows Explorer | Properties → Details → Remove Properties | Handles the common fields. Does not reach XMP or embedded image EXIF. |
The no-install route
Blackline rebuilds every page from rendered pixels and writes a fresh document, so the output carries no document information fields, no XMP packet inherited from the source, and no earlier revisions — the metadata is not stripped from the old file, it simply never exists in the new one. The same applies to images: the exported file is re-encoded from the canvas, so EXIF, GPS and embedded thumbnails do not come along. It runs entirely in your browser, which matters here: sending a file to a metadata-removal service means the service now has both the file and the metadata.
Check it afterwards. Always.
exiftool cleaned.pdf | grep -Ei "author|title|creator|producer|gps|date"
Expect nothing but the file size and format lines. If a name, a path or a coordinate is still there, the file is not ready to send.
Related: why the black boxes themselves leak, and the pre-release checklist.