Inspect metadata signatures, create a fresh output file with Deep Clean, verify the result and optionally add only the SEO metadata you choose.
The built-in scanner looks for common EXIF, XMP, IPTC/Photoshop, JPEG comment, PNG text, WebP metadata, C2PA/JUMBF indicators and selected AI-related textual markers when they are present in detectable file structures.
Structural information needed to display an image, such as ordinary image data and some color-management information, is treated separately from removable privacy metadata.
Deep Clean decodes the visible image and writes a fresh image file. This is the recommended privacy-cleaning mode because original metadata containers are not simply copied to the output.
Lossless Strip keeps the original encoded image data and removes known metadata blocks. It is faster but less defensive against unusual or unsupported structures.
Select a supported image and review what the scanner detects before making changes.
Use Deep Clean for a fresh re-encoded file or Lossless Strip when preserving encoded pixels matters.
The new file is scanned again and the before/after removable-marker counts are shown.
Removing file metadata does not guarantee that Instagram, Facebook or another platform will stop identifying content as AI-generated. A platform may use invisible watermarks, perceptual fingerprints, remote provenance matching or its own content classifiers.