C2PA Checker
Drop an image or video to read its Content Credentials — whether it was AI-generated, which model made it, and who signed it.
This tool reads what a file claims about itself. It does not cryptographically verify the signature or check the signer against a trust list, so a determined forger could write a manifest saying anything. Treat the result as a strong hint, not proof. For a full cryptographic verification, use the official Content Authenticity verifier. Equally, a file with no credentials is not necessarily AI-free — most tools strip or never add them.
How it works
Three steps, all of them on your own machine.
Drop your file
Pick an image or video, or paste one from your clipboard. Nothing is uploaded — the browser reads the bytes directly off your disk.
We read the manifest
Content Credentials live in a JUMBF block tucked inside the file header. We walk straight to it and decode the claim, the actions and the signing certificate.
Get a plain answer
Instead of a wall of JSON you get the part that matters: was this made by AI, by which model, and who put their name to it.
Why use this checker
Built to answer one question fast, without shipping your files anywhere.
Nothing leaves your browser
There is no upload endpoint. The file is parsed locally, so private footage and unreleased work stay on your machine.
Milliseconds, not seconds
Credentials sit in the file header, so we read a few kilobytes rather than the whole file. A 500 MB video checks as fast as a thumbnail.
Answers the AI question
We surface the IPTC digital source type and any AI-disclosure assertion, so you see the model name rather than raw field codes.
No plugin, no wasm download
The parser is a few kilobytes of JavaScript rather than a multi-megabyte WebAssembly bundle, so the page is usable the moment it loads.
Common use cases
Where reading Content Credentials actually settles an argument.
Checking a source image
A photo lands in your inbox and you need to know where it came from before it goes out. If the file carries credentials, you will see the capture device or the model that generated it, plus who signed the claim.
Used by journalists, fact-checkers, and editors.
Vetting supplier assets
Marketplaces and ad platforms increasingly require AI disclosure. Drop a batch of creatives in and see at a glance which ones declare themselves as generated content.
Used by marketplace ops, ad reviewers, and brand teams.
Confirming your own exports
Generative tools attach credentials silently. Before you publish, check what your export is actually declaring about the model you used and the account that signed it.
Used by designers, video editors, and agencies.
Perfect for
Who gets the most out of a local Content Credentials reader.
Verify provenance on deadline without sending an unpublished image to a third-party server.
Photo editors, verification desks, producers.
Triage uploads at speed and spot the ones that openly declare themselves as AI-generated.
Moderators, policy analysts, platform ops.
Know exactly what metadata your renders carry before a client or a stock library sees them.
Art directors, retouchers, motion designers.
Inspect what your pipeline writes into files while you are building C2PA support.
Backend engineers, ML engineers, tooling authors.
Frequently asked questions
What are C2PA Content Credentials?
C2PA is an open standard for recording where a piece of media came from. Cameras and AI tools can attach a signed manifest listing what created the file and what was done to it. Adobe markets the same thing as "Content Credentials".
Does a missing credential mean the file is not AI-generated?
No. Most AI tools still do not attach credentials, and screenshotting, re-encoding or uploading to a social platform usually strips them. Absence of credentials tells you very little; presence tells you a lot.
Can Content Credentials be faked?
The manifest is signed, so tampering breaks the signature — but only if you check it. This tool reads the manifest without verifying the certificate chain, which is why we describe results as claims. For an adversarial case, verify with the official Content Authenticity tool.
Which file types are supported?
JPEG, PNG, WebP and ISO-BMFF media — MP4, MOV, AVIF and HEIC. GIF and WAV are recognised but their credentials are not fully decoded.
Is my file uploaded anywhere?
No. Everything runs in your browser using the File API. You can confirm it by opening your network tab, or by disconnecting from the internet after the page loads — the checker keeps working.