C2PA Checker

Read the Content Credentials inside an image: which camera or AI tool made it, what edits were recorded, who signed the claim and whether the signature is valid. The C2PA checker runs in your browser.

What the C2PA checker shows

When an image carries Content Credentials, the checker verifies the manifest and lists the parts that matter for deciding where the image came from:

Generator: the app or library that wrote the claim, such as ChatGPT, Adobe Photoshop, a camera firmware or Google's C2PA library.
Signed by: the organisation named in the signing certificate. This is who stands behind the claim.
Signed at: the time stamp of the signature, when one was recorded.
Actions: what happened to the image, for example created, opened, edited, resized, converted, with the software agent for each step.
Digital source type: whether the content was captured by a camera, generated by a trained model ("trained algorithmic media"), or is a composite that includes AI-generated parts.
Validation: whether the signature and the hashes still match the file. A broken signature means the file changed after signing.

How to use the C2PA checker

Drop the original image file onto the box, or choose it from your device. Screenshots never contain credentials.
Wait a moment while the C2PA reader loads (about 9 MB the first time, then cached).
Read the "Content Credentials" line: AI recorded, present without an AI step, or none found.
The details table below the verdict lists the generator, signer, time and every recorded action.

Reading C2PA results for AI images

trainedAlgorithmicMedia

This IPTC digital source type means a generative model produced the content. ChatGPT, Gemini, Firefly and Designer images use it. If the manifest validates, the image came out of that generator, and the C2PA checker reports it as AI-generated with embedded evidence.

compositeWithTrainedAlgorithmicMedia

Part of the image is AI generated, for example a photo with a generative fill or an expanded background. The original capture may be real, but the result is not a straight photograph.

digitalCapture

The content was captured by a camera. Combined with a valid signature from a camera maker, this is the best evidence of a real photo that exists today.

No digital source type

Many editors record actions such as "opened" and "color adjustments" without saying where the content came from. The manifest then tells you about the editing history, but not about AI.

Why most images have no Content Credentials

C2PA is still new, and the chain is fragile. Credentials are lost when:

the image is uploaded to a social network that strips metadata;
someone takes a screenshot or screen recording;
the file is converted or optimised by a tool that does not support C2PA;
the image is sent through most chat apps.

Some platforms read the credentials before removing them and show a label such as "AI info" instead. If a picture has no manifest, the other checks in this tool still apply: metadata tags and the detection model. The AI image checker explains how the three results combine.

C2PA, watermarks and SynthID

Content Credentials are metadata: a signed label stored next to the pixels. Invisible watermarks such as Google's SynthID are hidden inside the pixels and survive screenshots and recompression better, but they can only be read with the owner's detector. This C2PA checker reads the open standard; it cannot read SynthID. Images from Google tools often carry both, so a manifest found here and a SynthID check in Google's own apps can confirm each other.

For creators and newsrooms

If you publish photos and want people to trust them, sign them. Several cameras can sign at capture, Photoshop and Lightroom can attach credentials on export, and the open-source c2pa tools can sign files in a publishing pipeline. Readers can then verify your images with any C2PA checker, including this one, without trusting a black-box score.

C2PA checker FAQ

What is C2PA?

C2PA (Coalition for Content Provenance and Authenticity) is an open standard for attaching a signed history to media files. The record, called Content Credentials or a manifest, says which device or software created the file, what edits were made and who vouches for that claim.

Which AI tools add Content Credentials?

OpenAI adds them to images from ChatGPT and its image API, Google to images from Gemini and Imagen, and Adobe to Firefly and many Photoshop exports. Microsoft Designer and Bing Image Creator, several camera makers (Leica, Sony, Nikon, Canon on selected models) and some phones also sign their output. The list grows every few months.

Why does the C2PA checker say "None found" for an AI image?

Content Credentials are stored in the file, and most social networks, chat apps and screenshot tools remove them. An AI image re-saved or reposted usually has no manifest left. "None found" means the file carries no record, not that the image is real.

Can Content Credentials be faked?

The manifest is signed with a certificate, and any change to the image or the record breaks the signature, which the checker reports. Someone can still sign a misleading claim with their own certificate, so look at who signed it, not only that it is signed.

Is this the same as the Adobe Content Credentials inspector?

It reads the same C2PA standard with the open-source c2pa library, but it runs entirely in your browser and adds metadata checks and an AI detection model. Nothing is uploaded.