C2PA (Coalition for Content Provenance and Authenticity) is a standard that embeds "content credentials" into digital files. If you create or edit images with Adobe, Microsoft, or many AI tools, your files likely contain C2PA metadata — and that's what's triggering AI labels on Instagram, Facebook, Pinterest, and other platforms.
Here's what you need to know.
What is C2PA?
C2PA is a specification developed by the Coalition for Content Provenance and Authenticity, whose members include Adobe, Microsoft, Intel, and others. It defines a way to store a cryptographically signed "manifest" inside image and video files that records:
- Creation history — Which tools and services were used
- Modifications — Edits, AI generations, and transformations
- Assertions — Claims about the content (e.g., "AI-generated" or "human-created")
The goal is transparency: anyone can inspect the manifest to understand how the content was produced. For AI-generated content, this often means the manifest explicitly states that AI was involved.
How C2PA Gets Into Your Images
When you use tools that support C2PA, they write a manifest into the file:
- Adobe Photoshop (Generative Fill, Neural Filters, etc.) — Adds C2PA when AI features are used
- Adobe Firefly — Embeds C2PA in all generated images
- Microsoft Designer / Copilot — Uses C2PA for generated content
- Content Authenticity Initiative (CAI) partners — Many cameras and software now support C2PA
The manifest is stored in a dedicated block within the file format (e.g., in JPEG, it can live in an APP11 segment or XMP). It's invisible when you view the image but readable by software that knows how to parse it.
Why Platforms Use C2PA for AI Labels
Social platforms want to flag AI-generated or AI-edited content. C2PA provides a standardized, machine-readable way to detect it. Instead of guessing from pixels or heuristics, they can:
- Read the C2PA manifest
- Check for AI-related assertions
- Apply labels (e.g., "AI Info", "Made with AI") accordingly
This is why stripping C2PA metadata prevents the label — without the manifest, the platform has no structured signal to act on.
C2PA vs. XMP and EXIF
C2PA is not the only metadata that triggers AI labels. Many tools also write:
- XMP — Custom fields for prompts, seeds, model names
- EXIF — Software and camera info that may indicate AI tools
A complete metadata removal process should strip C2PA, XMP, and relevant EXIF fields. Tools like Remove AI Label handle all of these.
Technical Details: Manifest Structure
A C2PA manifest typically includes:
- Claim generator — The tool that created the manifest (e.g., "Adobe Photoshop 25.0")
- Assertions — Structured claims such as "c2pa.actions" (list of actions like "c2pa.ai.generated")
- Signatures — Cryptographic signatures for tamper detection
Platforms look for assertions that indicate AI involvement and use them to apply labels.
How to Remove C2PA Metadata
To prevent AI labels, you need to remove the C2PA manifest (and other AI-related metadata) from your file before uploading. This can be done with:
- Dedicated metadata removal tools — Designed to strip C2PA, XMP, and EXIF
- Re-export with metadata disabled — Some editors allow this; results vary
- Screenshot or re-capture — Creates a new file without the original manifest (with potential quality loss)
For technical users and creators who need reliable, quality-preserving removal, a tool that specifically targets C2PA and related metadata is the most effective approach.
Conclusion
C2PA is a powerful standard for content provenance, but it's also what enables platforms to automatically label AI-generated content. Understanding how it works helps you make informed decisions about metadata removal. If you want to avoid AI labels, strip the C2PA manifest and related metadata before uploading — Remove AI Label can do this for you in seconds.