🖼

Drop your image here or click to browse

Supports PNG, JPG, WebP • Max 10 MB

What Is C2PA Metadata?

C2PA stands for the Coalition for Content Provenance and Authenticity, an industry standard for embedding provenance information into digital media. When Claude generates an image, it may include C2PA metadata, also known as Content Credentials, that records the image's origin as AI-generated, the model and version used, the timestamp of creation, and other provenance details.

This metadata is embedded within the image file itself, typically in the EXIF, XMP, or a dedicated C2PA manifest store. It's invisible when viewing the image but can be read by compatible software, social media platforms, and content verification tools. Platforms like Adobe's Content Authenticity Initiative, Google, and various social networks are increasingly checking for this metadata to label AI-generated content.

How Image Watermark Removal Works

Unlike text watermarks, which require careful token substitution, image watermark removal is conceptually simpler. The C2PA metadata resides in the file's metadata layers, completely separate from the actual pixel data. Our tool strips these metadata fields while leaving every pixel of the image completely unchanged, the visual content is identical before and after processing.

Specifically, our tool removes the following from image files:

  • C2PA manifest store: The primary container for Content Credentials, including claims, assertions, and signature data
  • XMP metadata: Extensible Metadata Platform fields that may contain AI provenance tags
  • EXIF AI markers: Any EXIF tags indicating AI generation, including custom Anthropic fields
  • IPTC metadata: International Press Telecommunications Council fields referencing AI creation

What About Invisible Pixel Watermarks?

Some AI image generators embed watermarks directly into the pixel data using steganographic techniques, subtle patterns invisible to the human eye but detectable by algorithms. As of current information, Claude's image generation primarily uses C2PA metadata rather than pixel-level steganography for provenance tracking.

If pixel-level watermarking is present, removal requires a different approach: subtle image transformations like re-encoding, minimal noise addition, or format conversion that disrupt the steganographic pattern. Our tool focuses on metadata removal as the primary and most reliable method for Claude-generated images.

Important Considerations

Legal note: Removing content provenance metadata may be subject to regulations in some jurisdictions. The EU AI Act and other emerging legislation may require disclosure of AI-generated content. Use this tool responsibly and in compliance with applicable laws.

All image processing happens entirely in your browser using client-side JavaScript. Your images are never uploaded to any server. The original file is read locally, the metadata is stripped in memory, and the clean file is generated for download, all without any network requests.

We recommend keeping a copy of the original image with its metadata intact for your records. While our tool produces a clean copy, having the original ensures you can always prove provenance if needed.

Working with Text Instead?

Our text tool handles Claude's statistical token watermarks.

✨ Remove Text Watermark