ChatGPT Image Watermark Remover

A free AI image watermark remover that detects and strips invisible C2PA Content Credentials from ChatGPT, DALL·E and Sora images. Runs in your browser: no upload, no signup.

AI Image Watermark Remover (C2PA Cleaner)

Drop images here, or click to upload

Up to 50MB per file · multiple files · JPEG · PNG · WebP

What are ChatGPT image watermarks?

Images from ChatGPT, DALL·E and Sora carry an invisible C2PA (Content Credentials) manifest: signed metadata describing the tool that made the file. Platforms can read it and show a "Made with AI" label. It is stored as an APP11/JUMBF segment in JPEG, a caBX chunk in PNG and a C2PA chunk in WebP.

How the AI image watermark remover works

  1. Scan: the file is checked for C2PA markers (JPEG APP11/JUMBF, PNG caBX, WebP C2PA).
  2. Re-encode: pixels are redrawn on a canvas, dropping C2PA, EXIF, XMP and ICC data.
  3. Download: save each file with a _cleaned suffix, or everything as a zip.
exiftool -a -G1 file_cleaned.jpg | grep -i -E 'jumbf|c2pa'

Limits: this removes metadata only. Visible marks and pixel-level watermarks such as SynthID are not removed. Do not use it to deceive people or evade disclosure rules that apply to you.

Frequently asked questions

Are my images uploaded?

No. Everything runs locally in your browser using the Canvas API.

Does it remove SynthID or visible watermarks?

No. It only strips metadata such as C2PA, EXIF and XMP.

Will image quality change?

Pixels are preserved. PNG stays lossless; JPEG and WebP are re-encoded at 95% quality.