Remove AI Metadata
Strip C2PA Content Credentials, EXIF, XMP and AI generator tags so crawlers and social platforms stop flagging your images as AI-generated.
Drop your images here
or click to browse. Supports JPG, PNG, GIF, WEBP, SVG
Remove AI Metadata — free, in your browser, no signup
Every major AI generator now embeds provenance data in your exports — C2PA Content Credentials, EXIF and XMP tags, ICC profiles, PNG text chunks and generator names like DALL·E, Imagen or Firefly. That is what Meta, LinkedIn, X and image crawlers read when they slap an “AI info” label on your post. This tool re-encodes your image from raw pixels in your browser, so every one of those containers is dropped deterministically — and tells you exactly which signals it found. Optional resample and dither passes degrade fragile pixel-domain watermarks on a best-effort basis. Nothing is uploaded anywhere.
Upload your image
Drag & drop, or click to browse — supports JPG, PNG, WebP, GIF, SVG
Configure options
Pick presets or fine-tune settings to match your exact use case
Download instantly
Export your result — no watermark, no signup, no upload to a server
Common use cases
- ✓ Remove the C2PA Content Credentials OpenAI attaches to DALL·E exports
- ✓ Strip EXIF, XMP and generator tags before posting to Meta, LinkedIn or X
- ✓ Clean AI provenance metadata from images you have edited and now own
- ✓ Remove GPS and camera EXIF from ordinary photos for privacy
- ✓ Normalise a whole batch of assets to one clean, metadata-free format
Frequently asked questions
What metadata does this actually remove?
All of it. Because the image is rebuilt from decoded pixels onto a fresh canvas, C2PA Content Credentials, EXIF, XMP, ICC profiles, IPTC blocks and PNG tEXt/iTXt chunks cannot survive the re-encode. The output contains only pixel data.
Will Meta or LinkedIn still label my image as AI-generated?
Those platforms read embedded C2PA / IPTC provenance fields to apply their “AI info” labels. Removing those fields removes the signal they rely on. Platforms may still apply labels from their own classifiers or from your account history, which no file-level tool can control.
Can it remove SynthID?
Not deterministically, and we will not pretend otherwise. SynthID is a statistical signal woven into pixel values. The Balanced and Maximum strengths resample and dither the pixel grid, which degrades fragile pixel-domain watermarks best-effort. Metadata removal is guaranteed; pixel watermark removal is not.
Does it change how my image looks?
Light mode touches metadata only. Balanced applies a gentle resample and ±1-level dither that is visually identical. Maximum uses a stronger resample plus a 2px edge trim, which is still invisible in normal viewing.
Is my image uploaded to a server?
No. Detection and re-encoding both run in your browser with the File and Canvas APIs. Your files never leave your device.
Is removing AI metadata legal?
For images you generated or own, removing embedded metadata is generally lawful — it is your file. It does not make AI content human-made, and it does not exempt you from disclosure rules that apply to you, such as academic, advertising or platform policies. Use it for privacy and control, not deception.