Embedded text and XMP
Prompts, comments, software names or document identifiers may be readable when an exporter wrote them into the file.
Midjourney / publishing privacy
Inspect the actual downloaded JPG, PNG or WebP for readable prompts, seeds, app fields and provenance. Remove only the supported sections you do not want to publish.
CONTEXT / EVIDENCE
Midjourney’s creation interface lets a creator copy the prompt, Job ID and seed, and creation details can also appear alongside an image on midjourney.com. That information belongs to the service and job history; it does not prove that every downloaded JPEG contains the same data in EXIF.
Re-exports, browser extensions and later editing apps can still add plain-text fields, XMP or a provenance manifest. ClearTags therefore inspects the exact publishing copy and removes only metadata found in supported container sections.
Prompts, comments, software names or document identifiers may be readable when an exporter wrote them into the file.
Prompt, Job ID and seed can remain available from Midjourney controls even when they are absent from the downloaded image.
Do not assume C2PA is present or absent from a Midjourney asset until the particular file has been inspected.
Use the exact upscaled or edited asset you intend to publish, not a screenshot of the gallery.
Check text chunks, EXIF, XMP, comments and C2PA separately.
Create a new copy without supported prompt-bearing text, XMP, EXIF or provenance blocks.
Review Midjourney account and creation visibility separately; changing a local file does not delete its job history.