Free image tools

Check an image for Content Credentials (C2PA)

Find out whether a JPEG or PNG carries a C2PA manifest, and read what it declares: the software that made the claim, the assertion labels and any sign of AI generation. This tool detects and displays. It does not validate signatures.

This tool needs JavaScript. It runs entirely in your browser; nothing is uploaded.

Private by design. Your files are processed by your own browser on your own device. They are not uploaded, stored or seen by Archive.London. Tool terms

How to use the Content Credentials (C2PA) Checker tool

  1. Choose a JPEG or PNG, or drop it on the page.
  2. See whether a C2PA manifest (JUMBF box) was found in the file.
  3. Read the listed details: claim generator, assertion labels and any AI-generation assertions.
  4. Follow the link to the official verifier if you need the signature validated.

To check an image for Content Credentials, open the C2PA Checker on Archive.London and drop in a JPEG or PNG. The tool looks for an embedded C2PA manifest and lists what it can read, including the claim generator and any AI-generation assertions. It does not verify signatures, so confirm important results in the official Content Credentials verifier.

How to read the result

There are three possible outcomes, and each needs careful interpretation.

No manifest found. This is the normal result for most images in circulation. It does not suggest a picture is fake, altered or AI-generated. Most cameras and editing programs do not write Content Credentials, and platforms that re-encode uploads often remove them.

Manifest found. The file contains provenance claims. The tool shows the claim generator, meaning the software or device that produced the claim, and the labels of the assertions inside it.

AI-generation assertion present. A manifest may declare that the content came from a generative model, typically through a digital source type such as trained algorithmic media.

In every case, what you see is what the file says about itself. Without signature validation it remains an unverified statement.

Detection is not verification

A C2PA manifest is designed to be tamper-evident. Its claim is digitally signed, and the assertions are bound to the image content by cryptographic hashes, so a validator can tell whether the picture or the claims were changed after signing and whether the signing certificate chains to a trusted issuer.

This tool does not perform those checks. It locates the manifest and decodes the readable parts. That is enough to answer the question does this file carry Content Credentials, and what do they claim?, and it is quick and private, since the file is read in your browser and never uploaded.

It cannot tell you whether the claims are authentic. For any decision that depends on that, such as publication or attribution, open the file in the official Content Credentials verifier, which validates the signature and trust chain.

Where the manifest lives: JUMBF, APP11 and caBX

C2PA is an open technical standard from the Coalition for Content Provenance and Authenticity. Content Credentials is the public-facing name for the same thing.

The data is packaged in JUMBF, the JPEG Universal Metadata Box Format, a container of nested boxes. In a JPEG file the JUMBF data is carried in APP11 marker segments, split across several segments when the manifest is large. In a PNG it sits in a dedicated chunk named caBX. This is entirely separate from EXIF, IPTC and XMP, which is why an ordinary metadata viewer may show nothing.

Because the manifest is extra data in the file, any process that redraws and re-saves the image without C2PA support will drop it. Screenshots, most format converters and many upload pipelines have this effect.

Provenance for photographs made before any of this existed

Content Credentials record what happened to a file from the moment a supporting device or program first signed it. They say nothing about the decades of photography that came earlier. For a film archive, provenance rests on older evidence: the original negatives and transparencies, their sleeves and labels, contact sheets, and publication records showing where a picture ran.

That physical chain is what distinguishes archival photographs from synthetic images, and it bears directly on questions about the origin of training data. The subject is covered on the AI training data licensing page, and longer pieces on provenance and metadata appear in the journal.

Frequently asked questions

Can this tool tell me whether an image is AI-generated?

Only in a limited sense. If the file contains a C2PA manifest with an assertion declaring AI generation, the tool reports it. Most AI-generated images in circulation carry no manifest, either because the generator did not add one or because it was stripped later. A result of no manifest found says nothing about how the picture was made.

Does the checker verify that Content Credentials are genuine?

No. It detects the manifest and displays what can be read from it. It does not validate the cryptographic signature, check the certificate against a trust list or confirm that the image matches the hashes in the manifest. Use the official Content Credentials verifier for that. Until a manifest has been validated, treat its contents as an unverified claim.

Why does my image show no Content Credentials when the camera or app supports them?

The feature may be switched off, or the manifest may have been removed after capture. Editing in software without C2PA support, exporting through a converter, taking a screenshot or uploading to a platform that re-encodes images will all usually discard it. Check the original file straight from the device or application before drawing conclusions.

Is my image uploaded when I check it?

No. The file is read by JavaScript in your browser and the search for a manifest happens on your own device. Nothing is sent to a server or stored, and the page works offline once loaded. If you go on to use an external verifier, that service has its own terms, so read them before submitting unpublished or confidential pictures.

Which file types can be checked?

JPEG and PNG, the two formats in which embedded manifests are most often encountered for still images. The C2PA standard also covers other image, video, audio and document formats, and allows manifests to be stored separately from the file, but those cases are outside this tool. For them, use the official verifier or a full C2PA software library.

Licensing the archive for AI and data projects

95,487 photographs from one photographer and one rights holder. Read the dataset card, search the 503-folder inventory, discuss a sample set under NDA, or arrange a first look as scanning proceeds.

See AI and data licensingWrite to the archive