C2PA Checker: Free Content Credentials Check for Images

The standard

What is C2PA and what are Content Credentials?

The short version: a signed label inside the file that says which tool made it and what was done to it.

C2PA, the Coalition for Content Provenance and Authenticity, publishes an open standard for attaching signed provenance data to images, video and audio. Content Credentials is the public name for that data.

A C2PA manifest is a block of data inside the file. It can record the tool that created the file, the edits made since, and whether AI was involved, and it is cryptographically signed so later changes can be spotted. The coalition was founded in 2021 by Adobe, Arm, BBC, Intel, Microsoft and Truepic, and AI image tools such as OpenAI's image models and Adobe Firefly now attach Content Credentials to what they make.

Credentials only help while they stay in the file. That is the catch, and it is why a C2PA check works best alongside a pixel-level AI check. The full specification is at c2pa.org, and the consumer side is explained at contentcredentials.org.

What it reads

What does the C2PA checker read?

Four places a file records where it came from, read in one pass, then matched against a table of AI generators and editing apps.

Source checkimage · original file

EXIF Softwareread

XMP CreatorToolread

PNG text chunksread

C2PA manifestread

Matched against the signature table, returned with an evidence level

The checker reads the C2PA manifest, the signed Content Credentials block, and the ordinary metadata fields that cameras and apps write when they save a picture: EXIF, XMP and PNG text chunks.

When any of them names a known tool, the result shows the tool and an evidence level. Strong means a signed C2PA manifest or an unambiguous software tag names the tool; moderate means a tool-specific field is present but unsigned; light means a weak hint such as a keyword; none means nothing was found. In our own 21-image test, the GPT Image 1 and Adobe Firefly files still carried C2PA Content Credentials. These are the tools in the current table.

AI generators the checker can name

Midjourney
DALL-E
OpenAI, ChatGPT and Sora
Anthropic Claude
Google Gemini
Google Imagen
Google Veo
Stable Diffusion
Stable Diffusion WebUI
ComfyUI
InvokeAI
NovelAI
Adobe Firefly
Flux (Black Forest Labs)
Leonardo AI
Ideogram
Runway
Kling AI
xAI Grok
Luma AI
Pika
Recraft
PhotoRoom
Picsart
Bria AI
SeaArt
Microsoft Designer, Bing Image Creator and Copilot
Meta AI

Editing tools are reported separately

Adobe Photoshop, Adobe Lightroom, GIMP, Canva and Affinity Photo are classified as editing tools, not generators. If the last tool to save a file was Photoshop, the source check says so and does not call the image AI generated.

So are screenshots

Screenshots and screen exports get their own class, because a screenshot of an AI image carries the screenshot tool's signature, not the generator's.

How to check

How do I check if an image has Content Credentials?

Three steps, about a minute.

01

Get the original file

Download the image itself, not a screenshot. If it came by message or email, save the attachment. Ask the sender for the original if you only have a copy.

02

Upload it above

Drop it into the free checker on this page. JPG, PNG, WebP, BMP, TIFF and HEIC, up to 4.5 MB, 10 checks a day with no signup.

03

Read both results

The source check shows any tool named by the C2PA manifest or metadata, with an evidence level. The verdict and AI score show what the pixels say.

Want the full signed record, edit by edit? The C2PA community's own Verify tool shows the complete manifest. Use this page for the fast answer and the AI score it does not give you.

C2PA vs SynthID vs AI detection

Three ways to check an image, and what each one misses.

Provenance labels, invisible watermarks and pixel analysis answer different questions. Use them together.

C2PA

Content Credentials

Signed metadata inside the file. Names the tool when present. Easily stripped by uploads, resizing and screenshots. This page reads it.

SynthID

Invisible watermarks

Google's watermark, woven into the pixels of images from its own models. Survives more edits, but only Google's tools can check it, and only for Google-made content. This page does not read SynthID.

AI score

Pixel analysis

A trained model reads the image itself, so it works with no metadata and on any generator. It gives a probability, not a signed record. This page runs it on every upload.

Comparing tools? See the AI image detectors that tested best, or read how AI image detectors work.

No credentials found

Why does an image have no Content Credentials?

Usually because they were stripped on the way to you. Missing credentials are normal, and they do not prove an image is real or fake.

Content Credentials live inside the file, so anything that rewrites the file can remove them: a social upload, a messaging app, a resize or a screenshot.

Most social platforms and messaging apps strip EXIF and XMP on upload. Resizing in an image editor rewrites the file. A screenshot creates a brand new file that knows nothing about the original. In each case the source check reports none, and the result panel says so plainly.

Three things follow. First, no Content Credentials does not mean the image is real; most AI images you meet online have already lost theirs. Second, the AI score reads pixels, not metadata, so it still applies. Third, when you can, ask for the original file, because that is where the evidence lives.

We tested this. On 20 October 2026 we took a C2PA-signed reference image published by the C2PA open-source project, then re-saved it as a JPEG, resized it to half, and re-encoded it as a PNG, the equivalent of a screenshot. All three copies lost the manifest entirely.

Many generators also never wrote C2PA in the first place, and many cameras and phones still do not. That is why this page pairs the credentials check with the AI image detector score instead of relying on either one.

FAQ

C2PA checker questions.

What is a C2PA checker? ＋

A C2PA checker reads the Content Credentials stored inside an image file. These are signed provenance records, defined by the C2PA open standard, that can say which tool created the file and what edits were made. This free checker reads the C2PA manifest plus EXIF, XMP and PNG metadata and names the AI generator or app when it is recorded.

How do I check if an image has Content Credentials? ＋

Upload the original image to the free checker on this page. It reads the C2PA manifest and other metadata and shows any AI generator or editing app it finds, with an evidence level, plus an AI score from the pixels. No signup is needed for up to 10 checks a day.

Does no Content Credentials mean the image is real? ＋

No. Social platforms, messaging apps, resizing and screenshots strip Content Credentials, and many tools never add them. A missing C2PA manifest says nothing about whether an image is real, which is why this page also gives an AI score from the pixels.

Which AI tools add C2PA Content Credentials? ＋

Adobe Firefly and OpenAI's image models attach Content Credentials to the images they make, and both kept them in our own test files. Support is growing across generators, cameras and editing apps, but it is not universal, so many AI images carry no C2PA data at all.

What is the difference between C2PA and SynthID? ＋

C2PA is signed metadata stored alongside the image, which anyone can read but which is easily stripped. SynthID is Google's invisible watermark embedded in the pixels of content from Google's own models, which only Google's tools can check. This checker reads C2PA and metadata, not SynthID.

Is the C2PA checker free? ＋

Yes. You get 10 checks a day with no account, for files up to 4.5 MB. A free account gives 50 scans a month across image, video and voice, with files up to 12 MB.

C2PA Checker.