AI labels and file metadata.
Know the difference.
Understand metadata-based AI tags in images and videos, visible watermarks, invisible content signals and platform-applied AI labels.
Image & video guide · Updated 4 November 2026
Four things people call an “AI label”
For an image or video, the phrase can mean a tag stored in its file, text visible in the content, an invisible signal, or a badge shown by a publishing service. These are different mechanisms.
| Type | Where it lives | What this cleaner does |
|---|---|---|
| Embedded metadata-based AI tag | Image or video information blocks | Omits original image blocks on export; clears recognized video blocks and supported fields. |
| Visible label or watermark | Image pixels or video frames | Visible content remains. |
| Invisible content signal | Pixels, frames or encoded streams | Outside metadata cleaning’s scope. |
| Platform-applied AI label | A social platform’s interface and systems | No guaranteed effect. |
Image cleaning and video cleaning differ
Images are rendered into new JPG, PNG or WebP files without copying their original embedded metadata. Videos keep the original encoded picture and sound while supported descriptive fields in MP4, MOV or M4V containers are cleared. Technical data needed for playback remains.
Why a clean scan is not an AI-generation verdict
Metadata can be added, changed or absent for many reasons. Finding a provenance block is not the same as validating its contents. Finding no supported blocks does not prove that an image or video was made without AI.
Choose the right workflow
Use the image and video AI label remover to clean supported file metadata, then inspect the download. Use the video cleaner for a video-only drop zone. Preserve attribution and follow relevant disclosure requirements.
Further reading
How metadata cleaning works · Video metadata guide · Scope and limitations
Choose a file. Keep your original.
Clean supported metadata locally, download a new copy and inspect the result.