Security and society / Context

Does a photograph with credentials tell the truth?

Knowing where a file comes from helps interpret it. It does not automatically make its content or caption true.

A Canon EOS 50D camera
Photographic camera. Archive image, 2017; this does not imply the device supports C2PA. Alexander Wang / darkleiv · CC0 1.0. Resized; photograph cropped.

A photograph can be real and still be used to tell a false story. An old scene only needs a caption presenting it as something that happened today. Trust in images therefore has more than one layer: the file, its provenance and its context.

C2PA proposes a mechanism for associating verifiable information about content creation and handling. Its principles distinguish that provenance from judgements about whether the material is good or bad. Verifying an association does not verify the entire story someone tells with the image. [C2PA]

What provenance can tell us

A history is valuable because it supplies information for better questions. Who declared an operation? Which part of the journey is documented? What does the credential being checked actually say? Those questions should not be replaced by a general impression of an “approved photo”.

Imagine a photograph of a public event republished years later. Even if the file retains valid information about its origin, the new caption could place it somewhere else or at another time. This example shows why we need to read the text accompanying a file, not just inspect the file itself.

No credentials does not mean false

C2PA’s FAQ explains that metadata can become separated from a file, including accidentally. Finding a file without that information is therefore not enough to conclude that it is false. Its history needs investigation; missing data should not be turned into proof of manipulation. [C2PA]

An archive photograph can remain a useful source if its provenance can be investigated in other ways. Conversely, a technical chain of data does not by itself resolve the meaning of a scene. Both questions deserve their own space: what do we know about the file, and what can we support about the event depicted?

An interface should also avoid oversimplifying the answer. From an editorial perspective, an indicator should help explain what was checked rather than suggest that every claim associated with the photograph has received independent validation.

How we would read an image with this information

We suggest starting with what the publication claims: person, place, date and event. Then check the original source and examine what the credential adds. If there is a gap, describing it is better than filling it with a conclusion the available information cannot support.

In this Journal, for example, a caption saying “archive photograph” separates the image from the date of the news. Attribution identifies the author and usage terms. Neither means we have verified C2PA credentials or that the photograph documents the specific announcement in the headline.

Our view is that provenance can strengthen verification when communicated precisely. The step to avoid is turning a technical check into a guarantee of truth. Knowing more about a file is useful because it lets us keep asking questions, not because it removes the need to ask them.

Sources and references

Prepared with AI assistance. Editorial policy and corrections.

Book an audit

  1. 1 Your details
  2. 2 Date and time

Let’s get to know you.

Your details are sent when you confirm.