Tech Behind ThingsHow the ordinary machinery actually works

Data & Privacy

Your photographs carry a second file inside them

Every image holds a block of information the camera wrote about itself, the moment and often the exact place, and it travels with the picture.

Close-up of a man with binary code projected on his face, symbolizing cybersecurity.
Photograph by cottonbro studio via Pexels
Editorial note. Independent reporting and analysis. Nothing here is sponsored or paid for. How we work.

Comparisons of image metadata usually pick a winner. This one picks the circumstances, which is more useful.

The difference in one place

  • Metadata records device, settings, timestamp and often coordinates.
  • Whether it survives sharing depends entirely on the service.
  • Embedded thumbnails have occasionally preserved an original before editing.

A block of description travelling with the picture

Most image formats reserve a dedicated block of space for structured information stored alongside the pixels themselves. Cameras write the make and model, lens details, exposure settings, orientation and a precise timestamp into that reserved space. The orientation field is why a photograph appears the right way up in one program and rotated in another.

None of this is visible when the image is displayed, and it survives copying, renaming and moving between drives. Editing software adds its own entries, including the application used, the version number and sometimes a history of the adjustments applied.

Where the location comes from

Phones write coordinates obtained from satellite positioning, network information or both, depending on what was available. The precision recorded is often finer than people expect, and it can identify a specific room within a building. Some cameras store a direction as well as a position, taken from the magnetometer at the moment of capture.

In practice, location is usually a permission that can be denied to the camera application without affecting anything else it does. Denying it after the fact does nothing to images already captured, since the data is embedded in each file.

What survives being shared

Many social platforms strip metadata when processing an upload, largely because they re-encode the image for their own purposes. Messaging applications vary, and some preserve everything when a file is sent as a document rather than as a photograph. Sending the original file by email, or through a file transfer service that does not re-encode it, typically preserves the block intact.

In practice, stripping by a platform protects the public copy while the platform itself retains whatever the upload contained. The only reliable assumption is that a file sent directly to somebody arrives with everything it was carrying.

Screenshots and documents keep their own records

A screenshot records the device and the moment it was taken, and on some systems the application that was in focus. Documents produced by office software store author names, organisation details, revision counts and editing time. Files created from a template inherit whatever the template's properties contained, including a previous author's name.

Portable documents preserve the producing application and often retain content that was visually covered rather than removed.

Redaction by drawing a black rectangle is the classic failure, since the text underneath remains selectable in the file.

The thumbnail that remembers too much

Formats commonly embed a small preview so a file browser can display an image without decoding the whole thing. Some editing tools historically updated the pixels while leaving the embedded preview showing the original scene. That produced cases where a crop intended to remove something was undone by extracting the preview from the same file.

Modern software generally regenerates previews correctly, and the failure mode is worth knowing when handling older files. The safest approach for anything sensitive is exporting a fresh file rather than saving over the original.

Firmware updates change this behaviour more often than hardware does.

Checking and removing it

Operating systems expose a properties or information panel showing the main fields, which is enough for a quick check. Dedicated tools display everything, including private manufacturer sections that ordinary viewers do not show at all.

Removing metadata is usually offered as a strip option when exporting, or as a separate remove properties action. Re-encoding an image through a screenshot removes almost everything and also discards quality, which is a crude but effective approach. Removing it entirely can be undesirable, since copyright and attribution fields live in the same structure.

Side by side

ConsiderationWhat it means in practice
A block of description travelling with the pictureMetadata records device, settings, timestamp and often coordinates.
Where the location comes fromWhether it survives sharing depends entirely on the service.
What survives being sharedEmbedded thumbnails have occasionally preserved an original before editing.

The takeaway

The picture is what you meant to send; the block beside it is what you actually sent.

The constraint is almost always physical, and marketing rarely mentions which one.

Questions readers ask

Does cropping a photo remove the location?

Not by itself. Cropping changes pixels while the metadata block is normally preserved unless the tool strips it deliberately.

Can I turn this off at the camera?

Location can usually be disabled through a permission or a camera setting. Device and exposure details are generally always written.

Data & Privacyphotographyprivacyfilesdata
Mikkel Aas
Editor, Tech Behind Things

Mikkel edits Tech Behind Things and has taken apart more devices than he has successfully reassembled.

Also by Mikkel Aas