How to Remove Content Credentials (C2PA) From Photos

Short answer: Content Credentials are a signed record, based on the C2PA standard, that some apps and cameras write into an image file. They can describe how the image was made, which tools edited it, and sometimes who made it. To remove content credentials before sharing, you need a copy of the file with that record taken out. Our free tool to remove AI content credentials does this in your browser, along with GPS, EXIF and other hidden metadata. It does not change how the image looks, and it does not remove invisible watermarks.

This guide explains what the credentials hold, why you might remove them, and what removal can and cannot do.

What C2PA Content Credentials are

C2PA is the Coalition for Content Provenance and Authenticity. On its site, c2pa.org describes an "open technical standard for publishers, creators and consumers" to "establish the origin and edits of digital content". It compares Content Credentials to "a nutrition label for digital content".

The C2PA explainer defines a few terms:

• A manifest is "the set of information about the provenance of an asset consisting of one or more assertions". The explainer says this is what is called a Content Credential.
• An assertion is "a data structure which represents a statement about the asset".
• The record is signed. The explainer says that if any part of the credential "is tampered with it will invalidate one or more hashes", so it is tamper-evident.

The explainer is also clear about what credentials do not do: "Provenance information alone cannot tell you whether the digital content is true, accurate or factual."

For background on where these credentials come from, see do AI images have metadata and content credentials in cameras.

What Content Credentials can contain

What is inside depends on the app or device that wrote it. According to the C2PA explainer and Adobe's documentation, a credential can hold:

• Actions: each step performed on the image, such as created or edited.
• Ingredients: other files that were combined into the image.
• Issuer and date: the organisation that signed the credential and when.
• App or device: the software or hardware used.
• AI tool used: the name of a generative AI model, if one was involved.
• Credit and identity: Adobe's documentation lists details such as a creator's name or social media handles.
• Standard metadata: the C2PA explainer says credentials work alongside IPTC, XMP and EXIF. The Content Authenticity Initiative's documentation shows an Exif assertion example with GPS latitude, longitude and altitude fields.

Not every credential has all of these. Identity and location are optional in the examples above. But when they are present, they travel with the file.

Where the credential sits in the file

A Content Credential is not part of the EXIF block. In a JPEG, EXIF lives in one segment of the file header (called APP1), while a C2PA manifest is stored in a separate segment (called APP11). That is why a tool that only clears EXIF or GPS can leave the credential in place, and why a tool that only removes the credential can leave your GPS behind. If you care about both, clean both.

Why someone may want to remove them before sharing

There are honest, everyday reasons to share an image without its full history.

• Privacy. A credential can name you, link your social accounts, or carry location data. You may be fine with that for a client but not for a public post.
• Workflow details. The edit history can show which apps, files or steps you used. Some people would rather not publish that.
• Mixed edits. A real photo with one small AI-assisted step can carry an AI label for the whole image. Some photographers would rather explain their process in their own words.
• Clean files. Some people strip all metadata from every file they share, as a habit.

Stripping credentials is a choice about what data you share. It is not a way to change what the image is.

What removal does not do

This part matters most. Removing the credential removes a record from the file. It does not change the picture, and it does not erase every trace of how the image was made.

• Invisible watermarks can remain. Some AI tools add a watermark into the image content itself. Google DeepMind says SynthID embeds watermarks that are "imperceptible to humans", and that its image watermark is "designed to stand up to modifications like cropping, adding filters, changing frame rates, or lossy compression". Removing a metadata record does not touch a watermark like that.
• The credential can be found again. The C2PA explainer says provenance can be removed, but also describes "soft bindings" that "can either be implemented via invisible watermarking or fingerprint lookup" and "allow the Content Credential to be discovered in cloud-hosted storage". So a copy of the credential may still exist outside your file.
• The image looks the same. If an image was made with AI, it still looks like it was made with AI. Removing a label does not change the content.
• Other copies are unchanged. The original file, cloud copies and anything already shared keep their credentials.
• Platforms may still label content. Platforms can use their own checks, so do not count on a missing credential to decide how a site labels an image.

Use it honestly

This tool is for privacy and for choosing what data leaves your device. It is not for passing off AI-generated images as real photos. If a client, publisher, contest, school or platform requires you to disclose AI use, removing the credential does not change that duty. Say what you used. Some publishers and agencies may also require you to keep credentials intact, so check before you strip a file you are delivering. Our guide on how to tell if a photo is AI-generated covers why provenance helps people trust images.

How to remove content credentials with GeoTag.world

1. Open the remove AI content credentials tool on GeoTag.world.
2. Drop in a photo. It accepts JPEG, PNG, HEIC and WebP. The file is processed in your browser and is not uploaded.
3. For a JPEG, the tool checks for a C2PA manifest and shows a badge if it finds one, next to the other metadata the photo carries, such as GPS, camera and copyright. The badge check works on JPEG files only. For PNG, WebP and HEIC the tool cannot show a badge first, but stripping still removes the credentials.
4. Strip the metadata and download the cleaned file.

Know what you get:

• It removes everything, not only the credential. The download drops the C2PA record along with GPS, camera details, EXIF, IPTC and XMP. If you need to keep a copyright line, add it back afterwards.
• JPEG is stripped at the byte level. The hidden segments are removed without recompressing the image, so the pixels are unchanged.
• PNG and WebP are re-encoded, which drops the credential and other metadata.
• HEIC is saved as a JPEG. Browsers cannot edit HEIC directly, so the photo is decoded and saved as a new JPEG file.
• No account is needed, and you can clean photos one after another.

How to check the result

1. For a JPEG, run the cleaned file back through the tool. No C2PA badge should appear. For PNG, WebP or HEIC the tool cannot show a badge, so rely on the next step.
2. Open it in an EXIF viewer and confirm the GPS and camera fields are empty.
3. Keep your original in a safe place if you may need to prove how the image was made later.

For a wider look at metadata cleaning, see how to remove EXIF data from photos. And when you want a copy without its credential, GPS or camera details, use the free content credentials remover in your browser.

Sources: C2PA 2.4 explainer, c2pa.org, Google DeepMind on SynthID.