Skip to content

Concept · Introductory

Image provenance and metadata

Every digital photo carries data beyond its pixels: EXIF metadata written by the camera, platform-added identifiers, and increasingly a signed provenance record under the C2PA standard. This data can reveal where and when a photo was taken, which device took it, and whether it was edited or generated. This entry explains what each layer holds, who strips it, and how it bears on finding and removing photos.

Updated November 6, 2026 · 4 min read · reviewed November 6, 2026

A face search works on pixels. Everything else attached to an image file is provenance: data about where the picture came from. It is useful to investigators, dangerous to the person photographed, and routinely destroyed by the platforms through which photos travel.

What is in EXIF metadata?

EXIF (Exchangeable Image File Format, standardised by JEITA, current version 3.0 in 2023) is a block of tags inside JPEG, HEIF and most raw files. A phone writes the capture time, camera model, lens, exposure, orientation and, if location services are on, GPS coordinates to a few metres. Editing software adds its name and history. Some cameras write a serial number. None of this is visible in the image; all of it is readable by any file inspector.

Metadata layers and what they can reveal
LayerWritten byRevealsSurvives upload to major platforms?
EXIFCamera, phone, editorTime, device, GPS, softwareNo: Instagram, Facebook, X, WhatsApp and most sites strip it
IPTC / XMPPhotographers, agencies, editorsCreator, copyright, captions, edit historyMostly no, though some photo services keep it
File name and dimensionsDevice, platformDevice defaults, platform resizingReplaced by platform naming
Platform identifiersThe hosting siteAccount, upload time (server-side, not in the file)Held by the platform, not the downloader
C2PA content credentialsCapable cameras, Adobe tools, some AI generatorsSigned capture and edit history, AI generation flagStripped by most platforms as of 2026; LinkedIn and some newsrooms display them

Who strips metadata, and why?

Social platforms remove EXIF on upload for two reasons: privacy (a GPS tag in a selfie gives away a home address) and bandwidth (the tags add bytes). Messaging apps do the same and recompress. The effect is that a photo downloaded from a profile has none of its original metadata, so the source page and the surrounding context are the only provenance left. Face search engines, including FaceSearch, strip metadata from uploads for the same privacy reason before running a search, a practice worth checking in each engine's policy.

What is C2PA and does it help?

C2PA content credentials are a 2021 open standard from the Coalition for Content Provenance and Authenticity (Adobe, Microsoft, Intel, BBC, Sony and others). A capable device or tool signs a manifest that records how the image was captured and each edit applied; AI generators that adopt it mark their output as generated. Verification tools show the chain. The limits are practical: the manifest is optional, it is removed by any platform that re-encodes the image, and absence of a credential proves nothing. In 2026 adoption is growing in cameras (Leica, Sony, Nikon models), Adobe products and some AI services; social platforms mostly strip it.

  1. Step 1

    Capture

    A camera writes EXIF; a C2PA camera also signs a manifest.

  2. Step 2

    Edit

    Software updates EXIF and, if C2PA-aware, appends a signed edit action.

  3. Step 3

    Upload

    The platform re-encodes the file; EXIF and usually the manifest are removed.

  4. Step 4

    Download or scrape

    The copy carries only pixels and the platform's file name.

  5. Step 5

    Trace

    Provenance now rests on reverse image search, the source page and archived copies.

How does provenance help when tracing a photo?

  • Original files with EXIF establish who took a photo and when, which supports a copyright claim under the DMCA and similar regimes.
  • Edit history (XMP or C2PA) shows whether a published image was altered from the original.
  • A C2PA generation flag identifies AI output when present, which no visual inspection can do reliably.
  • Server-side data held by a platform (upload time, account) can be obtained by lawful request even when the file carries nothing.

What should you do with your own photos?

  1. Turn off location tagging for the camera app, or strip GPS before sharing outside platforms that do it for you.
  2. Keep originals with full metadata offline; they are your proof of authorship.
  3. Use a C2PA-capable tool if you publish professionally, and know that most platforms will remove the credential.
  4. Check what a photo contains before sending it directly (email, cloud links, forums), since those paths preserve metadata.

Questions

Does a photo I post on Instagram reveal my location?

Not through EXIF; Instagram strips it. The location can still be revealed by what is visible in the image or by a location tag you add deliberately.

Can I see the metadata of a photo someone sent me?

If it came as a file (email, cloud link), yes, with any EXIF viewer. If it came through a messaging or social app, the metadata is usually gone.

Do face search engines use metadata?

Not for matching; they compare faces. Several state that they strip metadata from uploads before searching.

Is a photo without content credentials fake?

No. Most real photos in circulation have no credentials. The flag only helps when present.

Sources

  1. 1.C2PA, Content Credentials technical specification (checked January 15, 2025)
  2. 2.JEITA, Exif 3.0 standard (CP-3451F), 2023 (checked May 1, 2023)
  3. 3.Meta, How we handle metadata on photos (Help Centre) (checked November 6, 2026)

Entry reviewed November 6, 2026. Found an error? Tell us.

Cite this entry

FaceSearch Wiki editors. "Image provenance and metadata". FaceSearch Wiki, updated November 6, 2026. https://facesearch.wiki/concepts/image-provenance-and-metadata/