AI image metadata and C2PA, explained
AI-generated images can carry metadata that says how they were made. Some of it is ordinary EXIF-style data; some is signed provenance called C2PA Content Credentials.
Updated 2026-11-12
Direct answer: AI image metadata is information stored in an image file about how it was made — for example a generator or software name, a prompt in a text field, or signed C2PA Content Credentials. Ordinary EXIF can be viewed and removed with a metadata tool; C2PA is a separate signed record.
What is C2PA?
C2PA (Coalition for Content Provenance and Authenticity) is an open standard for attaching a signed history to a file — who or what created it and which edits were made. The user-facing name is Content Credentials. It is stored inside the file, but it is not the same as EXIF.
Where else can AI info appear?
- Software/Make fields in EXIF naming the tool that produced or saved the file.
- PNG text chunks and XMP where some generators write settings or prompts.
- C2PA manifests when the generator or editor supports Content Credentials.
Does removing EXIF remove C2PA?
Not necessarily as a promise. Re-encoding or stripping a file can remove embedded credentials, but that depends on the tool and the file. Metadata can also be missing simply because a platform stripped it, so its absence does not prove an image is real or fake.
What EXIF+ does and does not do
EXIF+ reads and edits EXIF, IPTC and GPS on your iPhone. It does not verify C2PA signatures or detect AI images. To see which basic fields an image carries, use the image metadata viewer.