Disclaimer and limits
What it does
- Removes C2PA, XMP, EXIF, IPTC and PNG text chunks from JPG / PNG / WebP.
- Removes udta, meta and iTunes-style tags at the MP4 / MOV container layer, without re-encoding.
- Removes the PDF Document Info dictionary and the XMP metadata stream.
- For HEIC / AVIF, decodes and re-encodes, stating clearly that the format changes.
- Everything runs in the browser; files pass through no server.
What it cannot do
| Item | Why it cannot |
|---|---|
| Pixel-level invisible watermarks (SynthID and similar) | Fused with pixel data; removing it means regenerating the image |
| Platform visual classifier decisions | The decision happens on the platform's servers |
| History retained on the platform side | Lives in the platform's database, unrelated to the file |
| Guaranteeing platform policy stays the same | Detection logic can change at any time |
| Fixing copies already in circulation | Metadata lives in every copy; each has to be processed |
| RAW formats (CR3 / NEF / ARW) | Structurally too different; not supported |
Being flagged after cleaning is normal
If a platform still shows an AI label after cleaning, the trigger was never in the metadata layer. There are three common causes: a pixel watermark, the platform's own visual classifier, and the platform's history for your account or that file. None of them is within reach of a metadata tool, and cleaning again will not help.
What you can consider instead: use different material, declare the AI assistance inside the platform, or appeal to the platform. This tool will not and cannot tell you why something was labelled.
Your responsibility
This tool is for people who hold the rights to their content. Uploading someone else's work and stripping its metadata may infringe copyright and moral rights, and may violate platform terms. That responsibility rests entirely with the user.
- Regulations such as the EU AI Act require disclosure of AI-generated content in certain situations; removing metadata does not remove that obligation.
- Regulated fields — advertising, news, healthcare, finance — have their own rules. Consult a professional.
- Some platform terms forbid removing provenance information; doing so may get an account penalised.
- Files kept as evidence, as archives, or for rights enforcement should not have their metadata removed.
Not legal advice
Everything on this site — page copy, comparison tables and FAQs alike — is technical explanation and does not constitute legal advice. For a specific compliance situation, consult a qualified professional.
Technical disclaimer
Cleaning works by rewriting the file structure. For files that are structurally unusual, damaged, or that use non-standard extensions, the rewrite may fail or produce unexpected results. Keep an original copy before processing anything important. The tool is provided “as is” and accepts no liability for loss arising from its use.
If a file will not open after cleaning, send a sample through thecontact page— that class of problem is a priority fix.