What each format carries
- JPEG: APP1 (EXIF, XMP) and APP13 segments, COM comments. Camera, lens, settings, date, often GPS.
- PNG: tEXt, iTXt, zTXt and eXIf chunks. Generating software, sometimes the complete prompt and generation parameters.
- PDF: XMP packet and Info dictionary. /Producer, /Creator, creation and modification dates, sometimes the source file path.
- DOCX and ODT: docProps/core.xml, app.xml, meta.xml. Author, last user who saved, cumulative editing time, revisions.
- MP4 and MOV: udta, meta and uuid XMP atoms. Device model, encoding software, dates, sometimes location.
- SVG and HTML: metadata blocks, RDF, meta tags and generator attributes.
Why it matters more than people think
A photo sent by direct message or published on a personal site — outside the social networks that strip metadata by default — can pin a home address to within a few metres. A PDF sent to a client can reveal it was produced from another client's folder. A DOCX can show an author who has nothing to do with the document any more.
In an AI context there is an extra risk: the metadata of a generated image regularly contains the model name and the prompt. Publishing the file as-is means publishing your workflow.
Cleaning without degrading
Stripping metadata recompresses nothing: image data is copied verbatim and only the side blocks disappear. Visual quality is intact and the file is lighter — sometimes by hundreds of kilobytes when an EXIF thumbnail was embedded.
One nuance to know: EXIF orientation. On some phone photos the correct rotation is described in metadata rather than applied to the pixels. A careless strip-everything pass can therefore leave the image sideways. A correct tool bakes the orientation into the pixels before dropping the field.
Local, batched, verifiable
These formats can be parsed in JavaScript: nothing requires you to upload files to a third-party service just to drop a metadata block — which would be self-defeating, since you would be handing over the complete file with its metadata intact.
Keep the round-trip habit: reload the cleaned files and the report should be empty. And remember that removing metadata removes neither a pixel watermark nor a C2PA manifest: those are three distinct layers.
Frequently asked questions
How do I remove GPS data from a photo?
By dropping the EXIF segment that holds it. The file is rewritten without that block, pixels unchanged, and reloading it in the analyser confirms no coordinates remain.
Don't social networks already strip metadata?
Most drop EXIF on upload, but not every block, not on every format, and not in files you send as an attachment or over messaging. Cleaning before you send is the only real control.
Can several files be cleaned at once?
Yes. Processing happens file by file inside the browser with no quota, and each file gets its own detection report.
Clean your file or your text
Detection and cleaning inside your browser: no upload, no account, no quota. Your files never leave your device.
Open the studio