Try now
AI Label Remover

How it works

AI disclosures can be triggered by file-level provenance, although platforms may use other signals too. Here is what cleaning removes from photos and videos.

The records inside media files

An image or video is a container. Alongside pixels, video frames and audio, it can carry descriptive data that a normal player does not show you:

  • C2PA content credentials — a cryptographically signed manifest describing how the file was made and edited, stored in a JPEG APP11/JUMBF segment, a PNG caBX chunk, a WebP C2PA chunk or an ISO-BMFF box in AVIF and HEIC.
  • XMP — an XML packet holding the generator name, prompt, seed, sampler and edit history, including the digitalSourceType field whose value trainedAlgorithmicMedia is a direct machine-readable declaration that an image was AI-generated.
  • EXIF — camera make and model, lens, exposure, timestamps, device serial numbers, GPS coordinates and a Software field that frequently names the AI tool.
  • IPTC — credit, source and attribution fields Adobe and agencies use for AI disclosure.
  • PNG text chunks — tEXt, iTXt and zTXt chunks where Automatic1111, Forge and ComfyUI store the entire generation workflow in plain text.
  • Video container records — MP4/MOV boxes and WebM elements can store titles, comments, encoder names, creation times, chapters, attached artwork and provenance manifests alongside the streams.

What the platforms actually read

Platforms can inspect provenance records at upload time and may combine them with invisible watermarks, perceptual matching, content classifiers and platform-side history. This explains why real media can receive a disclosure after an AI-assisted edit: a file can carry a machine-readable marker even when only a small part of the workflow used AI.

How image cleaning works

With the default Auto setting, JPG, PNG and WebP metadata sections are removed directly without recompressing the encoded picture. Dimensions, pixels and visual quality remain unchanged. AVIF and HEIC use a lossless PNG fallback because browsers do not expose a safe metadata-only rewrite for every variant.

How video cleaning works

A WebAssembly build of FFmpeg runs inside the browser and remuxes only the primary video and optional audio stream into a clean MP4 or WebM container. Stream copy means the encoded video and audio are never recompressed, resized or quality-reduced. Source metadata, chapters, attached thumbnails and extra streams are not copied. Files up to 200MB are supported.

The four steps

  1. Drop your media. The photo or video is read into browser memory with the File API. Its contents are not sent to an upload endpoint.
  2. Inspect what is inside. Images are inspected on the spot for C2PA, XMP, EXIF, IPTC and PNG-chunk records. Videos begin a local WebAssembly remuxing job.
  3. Clean without quality loss. JPG, PNG and WebP metadata is stripped without recompression. Video and audio streams are copied into a clean container without re-encoding.
  4. Download and post. You get a cleaned image or video download with original visual and audio quality. File-level metadata is removed, but watermarks and classifier signals may remain.

Privacy

There is no media upload endpoint in this application. Files are read with the File API, then processed with Canvas or FFmpeg WebAssembly inside your browser tab. The FFmpeg runtime is downloaded from this site when video processing first starts, but the selected video itself is not transmitted. Read the privacy policy.

Back to the tool