C2PA
In plain wordsAn industry group, and the technical standard it publishes, for attaching tamper-evident records of origin and editing history to digital photos, video, audio and documents.
The Coalition for Content Provenance and Authenticity, a standards body hosted by the Joint Development Foundation (part of the Linux Foundation family), and its open technical specification for content provenance. The specification defines manifests — signed bundles of assertions about a file's creation, edits and ingredients — bound to the content by cryptographic hashes and signed with certificates, so that tampering is detectable and the signer identifiable; manifests may be embedded in a file or stored remotely. Later versions added 'soft bindings', ways to link watermarksAI watermarking, n.: Hiding an invisible signal in AI-generated text, images, audio or video so that software can later tell the content was made by AI — like a banknote's watermark, but undetectable to the eye. or content fingerprints to a manifest so that stripped metadata can be recovered. Its user-facing form is Content CredentialsContent Credentials, prop. n.: A tamper-evident 'nutrition label' attached to a photo, video or document recording where it came from — which camera or AI tool made it and how it was edited — that anyone can inspect.. The standard is cited in policy debates on labeling AI-generated content, though it verifies provenance claims rather than truth.
Our image generator signs every output with a C2PA manifest stating that it was created by AI.
Origin
Founded in February 2021 by Adobe, Arm, the BBC, Intel, Microsoft and Truepic to merge the standards work of two earlier efforts, Adobe's Content Authenticity Initiative and Project Origin (led by Microsoft and the BBC), each of which carried on separately. Specification 1.0 appeared in January 2022; Google, OpenAI, Meta and Amazon joined its steering committee in 2024.
Related
- Content CredentialsContent Credentials, prop. n.: A tamper-evident 'nutrition label' attached to a photo, video or document recording where it came from — which camera or AI tool made it and how it was edited — that anyone can inspect.
- AI watermarkingAI watermarking, n.: Hiding an invisible signal in AI-generated text, images, audio or video so that software can later tell the content was made by AI — like a banknote's watermark, but undetectable to the eye.
- SynthIDSynthID, prop. n.: Google's system for hiding an invisible watermark in pictures, video, audio and text made by its AI tools, so they can later be recognized as AI-generated.
- data provenancedata provenance, n.: The documented history of a piece of data: where it came from, who created and changed it, under what permissions, and how it was processed before reaching an AI system.
- deepfake detectiondeepfake detection, n.: Using software, often itself AI, to tell whether a video, photo or voice recording has been faked or generated by AI rather than captured from reality.
Sources
- Coalition for Content Provenance and Authenticity (2022). C2PA Technical Specification 1.0.
Mentioned in
AI watermarkingAI watermarking, n.: Hiding an invisible signal in AI-generated text, images, audio or video so that software can later tell the content was made by AI — like a banknote's watermark, but undetectable to the eye., Content CredentialsContent Credentials, prop. n.: A tamper-evident 'nutrition label' attached to a photo, video or document recording where it came from — which camera or AI tool made it and how it was edited — that anyone can inspect., SynthIDSynthID, prop. n.: Google's system for hiding an invisible watermark in pictures, video, audio and text made by its AI tools, so they can later be recognized as AI-generated.