C2PA viewer for Content Credentials
Drop the image, then read the Content Credentials manifest if it is still there: who signed it, actions such as created or edited, and the tools named on those actions. That is what a verify view of the file can show. It does not list social-media download counts. Width and height can appear in an EXIF assertion, or you can read them from the file. A missing packet is inconclusive. Nothing is uploaded until you explicitly run the OpenAI check.
Definitions stay on their own pages: what C2PA is, Content Credentials, and content provenance. The short page is C2PA and Content Credentials. The head checker is the Content Credentials checker.
C2PA viewer for Content Credentials?
Drop the image, then read the Content Credentials manifest if it is still there: who signed it, actions such as created or edited, and the tools named on those actions. That is what a verify view of the file can show.
C2PA: Coalition for Content Provenance and Authenticity; C2PA specification 2.4; Instagram Help: label AI content before sharing; YouTube Help: disclosing use of GenAI content
Last updated
After you inspect a file, you can remove supported C2PA metadata. That removes supported container metadata only. It does not remove an AI label that a social app shows.
Read the definitions: what C2PA is, what Content Credentials are, how those two names differ, what content provenance means, how C2PA differs from EXIF, the AI provenance guide.
Forensic markers inspected by local CBOR decoder
What local inspection can read from container boxes:
- JUMBF box presence (JPEG APP11 0xFFEB, PNG caBX / c2pa chunks, WebP C2PA)
- Software Agent & Claim Generator (e.g. Grok Imagine, gpt-image, Adobe Firefly)
- DigitalSourceType (indicates declared source category such as trainedAlgorithmicMedia or digitalCapture)
- Action verbs (e.g. c2pa.created marks newly created asset in this action; c2pa.infilled or c2pa.edited marks localized editing)
- Claim signatures & manifest assertions
Strict Forensic Boundaries (What cannot be concluded):
- Absence of C2PA metadata does NOT prove an image is human-made (social platforms often strip headers)
- Presence of C2PA without trusted cryptographic hardware does not prove factual reality
- Local inspection parses binary structure; it is not a trusted platform authority validator
- Optional official OpenAI verification forwards original bytes to OpenAI only upon user click
Verify C2PA provenance yourself
Select an AI or camera photo
Choose an image generated by ChatGPT, Grok, or a C2PA-equipped camera (Leica, Sony).
Inspect manifest structure
Review extracted softwareAgent, claim_generator, and action verbs.
Examine actions and source
c2pa.created indicates an asset was newly created in that action; evaluate DigitalSourceType (trainedAlgorithmicMedia vs digitalCapture) and generator assertions to understand origins.
Optional official verification
If desired, click to run optional official OpenAI Content Provenance verification.
Methodology & Provenance Architecture
Our forensic engine uses a zero-dependency binary CBOR decoder to read assertions directly from JUMBF boxes. Read our full provenance guide and technical methodology.
Read AI Provenance GuideC2PA, Content Credentials, and content provenance
What can a Content Credentials verify tool show?
When the manifest is still in the file: the issuer, actions such as created or edited, and the software or tools named on those actions. Ingredients appear when the manifest lists them. Social-media download counts are not a C2PA field. Pixel width and height can sit in an EXIF assertion, or you read them by opening the file. A missing manifest is inconclusive. The longer walkthrough is the Content Credentials how-to.
What is C2PA?
C2PA is the Coalition for Content Provenance and Authenticity specification. It defines how a file can carry a signed manifest. This page reads that manifest when the file still has one. The longer definition is on the C2PA learn page. A missing packet is inconclusive.
What are Content Credentials?
Content Credentials are the name people use for that signed manifest: signer, software agent, and actions such as created or edited. This viewer reads them in the browser. It does not decide that a picture is real, and it does not invent an AI percentage.
What is content provenance?
Content provenance is the record of where a file says it came from and how it changed. C2PA is one way to store that record. This page shows the record when the file still has it. Absence does not prove a camera original or an AI image.
Does the viewer upload my image?
The local read stays in the browser. Official OpenAI verification forwards the original bytes only after you click it.
Will this remove an AI info label?
No. Viewing or even stripping a manifest does not control Instagram, Facebook, TikTok, or YouTube labels. Those products can also use self-disclosure and their own detection.