What you'll see in a result
- Manifest found / not found — whether the file carries Content Credentials at all.
- Signature state — Trusted (valid and on the C2PA trust list), Valid (cryptographically intact) or Invalid.
- Signer — the certificate holder that vouches for the record.
- Claim generator — the software that produced the manifest (an AI model, a camera, an editor).
- Digital source type — including trained algorithmic media, the standard’s term for AI-generated content.
- Actions and ingredients — recorded edit history and source files, when present, plus the full raw manifest behind “technical details”.
How C2PA differs from a watermark
Content Credentials are signed metadata that travel with the file — an evidence label, not a hidden mark. Anyone can verify them (that’s the point of an open standard), and any tampering with the file breaks the signature visibly. The trade-off: metadata can simply be stripped. Invisible watermarks like SynthID take the opposite trade — they survive edits better but only their provider can read them. Modern providers increasingly use both layers; the background is in our C2PA explainer.
Who attaches Content Credentials today
Anthropic (files generated by Claude), OpenAI (all ChatGPT/API/Codex images since May 2026), Adobe Firefly and Creative Cloud, Google products, and hardware from Leica, Sony and the Pixel 10 camera, among others. See the provider status table for the current picture.