AI image tools now stamp Content Credentials and older IPTC fields into files. That can be useful for provenance. It is a problem when you need a clean asset for a client who does not want model traces in the bytes.
I keep a dedicated inspector for this: Remove AI Meta.
What is actually in the file
Typical extras include C2PA manifests, IPTC software fields, and EXIF that still names the generator. Viewing the image is not the same as reading those chunks.
A publish checklist
- Inspect the download
- Strip credentials if the brief requires a clean file
- Re-inspect so you did not leave a second packet
- Keep the original if you still need provenance later
The before-and-after flow is on removeaimeta.com. Metadata is not the image. Treat it as a separate document attached to the pixels.
Top comments (0)