What this checker can verify
- Whether a C2PA manifest is present and whether its signature cryptographically validates against this exact file.
- Who signed it — the certificate issuer and, where present, the generating software (e.g. an AI model or an editing app).
- Declared origin — the digital source type, including “trained algorithmic media”, the C2PA way of saying AI-generated.
- Recorded history — actions, edits and ingredient files, when the manifest includes them.
What this checker cannot prove
Absence of Content Credentials proves nothing: most images never had them, and screenshots, edits and social platforms routinely strip metadata. Invisible pixel-level watermarks such as SynthID are separate from C2PA and can currently be verified only through Google’s own channels. A valid credential proves the provenance record is authentic and the file untampered — it does not verify claims the record never made.
Supported files
Tested at launch: JPG, PNG and WebP, up to 30 MB. Other image formats may work and are attempted when selected; results for untested formats are labeled accordingly if parsing fails.
Related
- C2PA checker — the same engine with a C2PA-focused explanation.
- ChatGPT watermark checker — OpenAI images carry C2PA + SynthID since May 2026.
- What are C2PA Content Credentials?