Skip to content

Metadata vs content provenance

These two terms get used interchangeably and should not be. The distinction matters for anyone reasoning about whether a file can be trusted.

6 min read · Updated

Three separate things

It helps to separate them explicitly:

  • Metadata — information stored inside the file: camera, settings, timestamps, location. Written by software, editable by software, unsigned.
  • Content provenance — cryptographically signed assertions about a file’s origin and editing history, attached by systems built for that purpose. Verifiable against a signing key.
  • Visual content — what the image or video actually depicts, which can be analysed independently of either.

Why metadata is not evidence

Metadata carries no signature and no chain of custody. Nothing about an EXIF field distinguishes one written by a camera from one written by a text editor five minutes ago.

That is not a flaw to be fixed; it is what metadata is. It was designed as a convenience for photographers, not as an attestation. Treating it as proof of origin is a category error.

What changing metadata does and does not achieve

Editing metadata changes what the file says about itself. It does not:

  • make a platform consider the file authentic;
  • make a platform consider it camera-originated;
  • prevent analysis of the image content itself;
  • remove or alter signed provenance assertions, which are a separate mechanism;
  • change other signals associated with the file or the act of uploading it.

Where this tool stands

This tool reads, removes and edits metadata fields, and converts files between formats. It says what it does plainly, and it does nothing else.

It does not implement anything designed to bypass provenance systems, defeat detection, or manufacture false evidence, and it makes no claim that a file processed here will be treated as authentic by anything. Those would be claims nobody can verify, and the product is more useful without them.

Why the confusion is common

Metadata looks authoritative. It is technical, it is precise to the second, and most people never see it, which lends it an air of being hidden truth rather than an ordinary editable field.

It also often is accurate, which makes the mistake easy. A camera writes a correct timestamp the overwhelming majority of the time. The problem is that "usually accurate" and "evidence" are not the same thing: a field that can be changed without leaving a trace cannot support a conclusion on its own.

Provenance systems exist precisely because that gap was worth closing. They add a signature, which is the part metadata never had.

The practical reading

If you are protecting your privacy, metadata removal is genuinely effective: it deletes the camera identity, the timestamps and the location that would otherwise travel with your file. That is a real outcome with a real benefit.

If you are trying to establish that a file is what it claims to be, metadata cannot help you in either direction — not by being present, and not by being absent. Look for a signed assertion, or corroborate the content itself.

Both things are true at once, and holding them together is the whole point: metadata is worth controlling and not worth trusting.

Questions

Can metadata prove when a photo was taken?

No. A timestamp records what the device’s clock said, assuming nothing changed it afterwards — and anything can change it afterwards. It is useful information, not evidence.

Does removing metadata make a file suspicious?

Not meaningfully. Most images in circulation carry no metadata, because re-encoding strips it. A file without metadata is indistinguishable from the very large number that never had any.