Invisible AI tags, removed
Your images.
No invisible tags.
When you make an image with Gemini or ChatGPT, it carries an invisible SynthID watermark and a C2PA record — and platforms read those to stamp your work “Made with AI”. ClearMark scrubs both out, right in your browser, so the file is properly yours again.
Clean a file
Drop an image, or click to pick one. PNG, JPG or WebP, up to 10 at once. Everything stays in your browser — nothing gets uploaded anywhere.
Drop images here or
or paste an image (Ctrl+V)
Before / After
Drag the slider. Both look the same — that is the point. The difference is what scanners can read.
Best-effort cleaning. Verify the result yourself at synthid.com (Google's free detector, ~10 checks/day).
What gets removed
SynthID signal
Google weaves this invisible watermark into the pixels of Gemini, Veo and Nano Banana images. Cropping or re-saving won't shift it — so we break it apart in the frequency domain, where it actually lives.
C2PA record
The signed “AI-made” tag attached to the file. We strip it out completely, along with EXIF and the rest of the metadata.
Generator metadata
Model names, prompts, tool fingerprints hiding in the file header. Gone — the file you download carries no record of where it came from.
How it works
- 1 · Drop your fileAn image from ChatGPT, Gemini, Nano Banana or any generator. It loads straight into your browser.
- 2 · We find the signalThe engine reads your image in the frequency domain and finds the watermark pattern hiding in the pixels.
- 3 · We rewrite it outSmall, targeted changes break the signal apart — while the picture looks exactly the same to your eyes.
- 4 · Download cleanA fresh file with no metadata, no C2PA tag, and a disrupted watermark — ready to post.
Honest note: no remover can promise 100%. SynthID is built to be stubborn, and detectors keep changing. ClearMark is best-effort by design — and we tell you how to verify it instead of asking you to trust us.