C2PA (Coalition for Provenance and Authenticity) is a technical standard for proving where a piece of media came from. Created by the Content Authenticity Initiative — founded by Adobe, Microsoft, Intel, Arm, BBC and others, now with hundreds of members including camera makers (Leica, Nikon, Sony), AI labs (OpenAI, Meta) and platforms (TikTok, YouTube) — it defines "Content Credentials": tamper-evident manifests embedded in files.

How a manifest works: when a C2PA-aware tool creates or edits an image, it writes a signed manifest into the file — JPEG APP11/JUMBF segments, PNG jumbd boxes, or XMP-adjacent stores in other formats. The manifest lists assertions (generator name, action history, timestamps, ingredient thumbnails) plus a cryptographic hash of the pixels and a signature chain back to the vendor's certificate. Change one pixel and the hash breaks — verifiers show "credentials invalid", which itself reveals tampering.

  • Who writes C2PA today: Adobe Firefly / Photoshop / Express, Microsoft Designer & Bing Image Creator, OpenAI DALL-E exports, Meta AI imagery, Leica M11-P / Nikon Z8-Z9 / Sony A9 III (in-camera signing), Canva.
  • Where you see it: Behance "Created with AI" badges, Instagram/TikTok "AI-generated" auto-labels, the verify.contentauthenticity.org inspector, and Photoshop's "cr" pin icon.
  • What it records about you: tool + version, edit chain, timestamps, and potentially Adobe/Microsoft account linkage and source-image thumbnails — more identifying than classic EXIF.
  • Why it survives normal cleaning: EXIF removers target TIFF/EXIF tags, not JUMBF boxes. C2PA needs structural segment removal at the byte level.
Privacy vs. provenance trade-off

C2PA was designed to fight deepfakes — a worthy goal. But for freelancers, sellers and everyday posters it means every AI-assisted draft is permanently fingerprinted and auto-labeled across platforms. If you need a neutral file (client delivery, stock upload, contest), strip the manifest after export. How-to: remove C2PA step by step; Adobe-specific: Firefly credentials guide.

How to check for C2PA: upload the file to Adobe's Content Authenticity Verify site, run exiftool -JUMBF -Provenance file.jpg, or simply drop it in the tool below — the preview explicitly reports "C2PA Content Credentials: Detected" when a manifest is present.

How to remove it: future files — disable the Content Credentials toggle in Adobe/Microsoft apps. Existing files — structural stripping: parse and drop APP11/JUMBF segments and PNG jumb boxes while preserving image data. That is exactly what the Vantawebs cleaner below does, losslessly and entirely in your browser.

Clean it now: drop the file below to detect + remove C2PA and all companion metadata in one pass.

Related: how to check if an image is AI-generated · does TikTok scan AI metadata?