Skip to content
Irregularpedia Logo Irregularpedia

Metadata Hygiene

Metadata Hygiene

Section titled “Metadata Hygiene”

Almost every file you create carries metadata — data about the data. A photo records the camera, the exact GPS coordinates, and the timestamp. A Word document records the author name, organization, edit history, and sometimes earlier deleted text. A PDF embeds the software that produced it. None of this is visible when you glance at the file, and all of it can deanonymize you, locate you, or contradict a cover story.

Metadata hygiene is the practice of knowing what your files reveal and stripping it before you share them.

Where metadata hides

Section titled “Where metadata hides”
SourceWhat it can reveal
Photos / video (EXIF, XMP, IPTC)GPS location, date/time, device make & serial, thumbnails of the original (un-cropped) image
Office documents (Word, Excel, PowerPoint)Author, organization, template path, revision history, comments, tracked changes
PDFsProducing software, author, creation/modification timestamps, sometimes layered original text under “redactions”
AudioEncoder, tags, sometimes device identifiers
Any file (filesystem)Created/modified timestamps, owner, on some systems extended attributes

Images and EXIF

Section titled “Images and EXIF”

Phone and camera photos embed EXIF tags. The most dangerous is the GPS geotag, which pins the photo to a precise latitude/longitude — enough to find a home or a safehouse. Cropping or blurring the visible image does not remove the embedded GPS data, and some formats also store an un-edited thumbnail of the original frame.

Inspect and strip EXIF with ExifTool:

Terminal window
exiftool photo.jpg                 # show all metadata
exiftool -gps:all= photo.jpg       # remove only GPS tags
exiftool -all= photo.jpg           # remove ALL metadata (writes photo.jpg, backs up original)

Documents: Office and PDF

Section titled “Documents: Office and PDF”

Office files are ZIP archives of XML and retain far more than the visible text — author, comments, tracked changes, and prior revisions.

  • Microsoft Office: File → Info → Check for Issues → Inspect Document, then remove document properties and personal information.
  • LibreOffice: Tools → Options → Security → “Remove personal information on saving.”
  • PDF: re-export or “sanitize” rather than trusting the source app. Adobe Acrobat has Redact → Sanitize Document; open-source options are below.

Redaction done right

Section titled “Redaction done right”

The classic failure is “redacting” a document by drawing a black box over text — the text is still selectable underneath. Redaction must remove the underlying content, not merely cover it. Flatten and re-rasterize, or use a tool that deletes the content stream. The U.S. NSA’s redaction guidance is a concise, authoritative reference.

Browser and network metadata

Section titled “Browser and network metadata”

Sharing happens over the network, which leaks its own metadata:

  • Referrer headers tell the destination which page you came from. Set a strict referrer policy (Referrer-Policy: no-referrer / strict-origin-when-cross-origin) on sites you control, and prefer browsers that minimize referrers by default.
  • Upload behavior varies by platform. Most large social networks re-encode images and strip EXIF on upload — but messaging apps, email attachments, file shares, and “original quality” uploads frequently do not. Never assume a platform scrubs for you; scrub first.

One-shot scrubbing tools

Section titled “One-shot scrubbing tools”
ToolUse
ExifToolRead/write metadata for images, PDFs, audio, video
mat2 (Metadata Anonymisation Toolkit)One-command stripping across many formats; mat2 file.pdf writes a cleaned copy
Platform “remove properties”Built-in, good for a quick pass but less thorough than the above
Terminal window
mat2 report.pdf        # produces report.cleaned.pdf with metadata removed
mat2 --show photo.jpg  # list what would be removed
Section titled “Related”