Sora C2PA & Content Credentials

Short answer

OpenAI documented that Sora-generated videos could carry C2PA Content Credentials — signed provenance metadata describing that the content was produced with Sora. When present and intact, that metadata can support origin claims in C2PA-aware readers. It is not a visible watermark, and it is not OpenAI’s internal tracing stack. AI Watermark Center’s inspector reads embedded Content Credentials in JPEG, PNG, and WebP still images. That capability does not imply a Sora video inspector, does not validate video containers, and must not be marketed as “Sora verification.” OpenAI also notes the Sora product is no longer available as of 26 April 2026; treat this as historical documentation.

What signed provenance can show

C2PA manifests can bind assertions about how an asset was created or processed, with cryptographic signatures when the implementation supports them. Finding a Sora-related assertion in a file (where a capable validator can read it) is evidence about documented provenance metadata — not a guarantee the depicted scene is real, and not a substitute for OpenAI’s own internal tools.

Why the image inspector is not a video inspector

Still-image inspection and video-container inspection are different engineering surfaces: codecs, metadata locations, fragment packaging, and performance constraints differ. Shipping JPEG/PNG/WebP C2PA support does not silently unlock MP4/WebM provenance parsing. This site does not offer fake Sora verification by reusing the image tool on video uploads.

Limitations

• No AWC video C2PA path.
• No claim that missing credentials proves non-Sora origin.
• Metadata can be stripped by re-encodes and some editors.

Related tools

Use the image inspector only for supported still formats — not as a Sora checker.

Inspect embedded Content Credentials (images)