Image tool
Use this browser tool as a C2PA checker and Content Credentials viewer for JPEG, PNG, and WebP. It reads EXIF, GPS where present, and embedded manifests — not SynthID, and not cloud-recovered credentials.
Loading the image inspector…
A still image is more than pixels. Cameras, phones, and editors may store capture time, device names, software history, captions, and sometimes precise GPS coordinates. Those fields travel with the file until something strips or overwrites them. Listing them is useful for privacy review and for understanding what evidence is actually present.
EXIF is a common way to store camera settings and capture details inside JPEG and some other formats. Make, model, orientation, exposure, and GPS IFDs are typical EXIF contents. Presence of EXIF is not a watermark, and missing EXIF is not proof that an image is synthetic.
Content Credentials are a product name for provenance information stored using the C2PA specification: signed manifests bound to a file. This inspector looks for an embedded manifest. It does not recover credentials from an external database, a thumbnail, or a soft binding.
When a manifest is present, the official CAI web SDK (@contentauth/c2pa-web 0.13.4, audited 2026-08-15) reports whether it could parse the store and whether it emitted validation status codes. The SDK types reference C2PA Technical Specification 2.3 structures. This inspector does not claim full C2PA 2.4 feature coverage. An empty error list is not a statement that the photograph is authentic or that its claims about the world are true. Trust-list verification is not enabled in this version, so a signature is not treated as a trusted identity.
No. Absence of embedded Content Credentials does not establish how an image was created. Manifests can be omitted, stripped, or stored in a form this inspector does not read. Invisible watermarks are a separate mechanism and are not tested here.
Only if the file actually contains a field or assertion that says so, and even then that statement belongs to the signer of those claims. This tool will surface AI-related digital source types or assertions when they appear in a readable manifest. It will not guess a model from EXIF software strings or from missing credentials.
No. SynthID and other invisible pixel watermarks are listed as not tested. Soft-binding recovery is not performed, and the file is not uploaded for matching.
Inspection runs in your browser. The selected image is not uploaded to AI Watermark Center for analysis. Analytics, if later enabled, will not include the image, filenames, coordinates, or metadata values. Files larger than 25 MB are rejected so the tab stays responsive. The C2PA WebAssembly module and worker are loaded from this site when you use the inspector.
Results follow the same evidence rules as the rest of the site. Read the research methodology for how documented, verified, observed, and unverified claims are labelled.
Related
Guide
What can actually be checked — routes to this inspector for C2PA/metadata.
Guide
Why a C2PA check is not a SynthID result.
Provider
Research cluster. This inspector is not a Nano Banana or SynthID detector.
Provider
Research cluster. This inspector reads embedded C2PA; it is not an OpenAI watermark detector and does not detect SynthID.
Guide
Signed provenance versus invisible watermarks.
Guide
Visible marks, credentials, hidden characters, and statistical systems.
Research
Official, verified, observed, and unverified claims.
Tool
Inspect hidden Unicode in pasted text.