What metadata does the cleaner remove?

Camera EXIF (make, model, lens, serial numbers, dates), GPS coordinates, IPTC and Photoshop segments, XMP packets, PNG text chunks including Stable Diffusion generation parameters, C2PA / Content Credentials manifests and JUMBF data, and PDF document info such as author, creator, producer and dates.

Are my files uploaded anywhere?

No. Inspection, cleaning and verification all run in your browser with JavaScript. No file bytes are ever sent over the network — the only network activity is an optional subscription check for the batch feature.

Cleaning one file at a time is completely free and needs no account. Processing several files at once or downloading them as a ZIP requires a subscription, since it's a batch convenience feature.

Does it remove C2PA Content Credentials?

Yes — JUMBF manifest boxes, the 'caBX' PNG chunk and C2PA chunks in WebP are removed, and the inspection step reports when Content Credentials were found. Note that pixel-level watermarking some generators add separately is not affected.

Can it fool AI detectors or remove SynthID?

No — and be careful with tools that claim otherwise. This tool removes container metadata. Watermarks woven into pixels and content-based AI detection do not depend on metadata and are not changed.

Will cleaning reduce image quality?

Not in lossless mode — the compressed pixel data is copied byte-for-byte. In re-encode mode the image is decoded and saved again at quality 0.95, which looks visually the same but re-compresses the pixels.

Privacy tool

AI Metadata Cleaner

See and remove the hidden data inside your images and PDFs — EXIF, GPS, XMP, C2PA Content Credentials and AI generation parameters — without uploading anything. Everything runs locally in your browser.

100% local processing

Files never leave your browser — no upload, no server.

JPEG, PNG, WebP or PDF · up to 25 files · 25 MB each · 150 MB total

Single files are free — no sign-in needed. Batch cleaning needs a subscription.

Cleaning mode

Re-encode & change fingerprintDecodes and re-encodes the image, removing every metadata container. Best against hidden metadata.Preserve original qualityRemoves metadata segments byte-by-byte without decoding or re-compressing. Pixels stay exactly the same.

PDF files are always cleaned without re-encoding: the Info dictionary and XMP packet are removed; images embedded inside the PDF are not rewritten.

Change image fingerprintApplies an invisible scale, rotation and noise perturbation so the new file no longer matches the original.

Queue

0/25 · 0 B

Add files to inspect and clean their metadata.

Deep metadata removal, verified

Every removable metadata container is stripped and the output is re-scanned to prove it.

Processed 100% locally

Files are read and cleaned inside your browser. Nothing is uploaded, queued on a server or logged.

Removes C2PA & provenance

Strips Content Credentials, JUMBF manifests and XMP provenance records embedded in JPEG, PNG and WebP files.

Strips AI generation parameters

Removes PNG text chunks such as Stable Diffusion / A1111 'parameters' prompts, plus EXIF, IPTC and Photoshop segments.

Lossless or re-encode

Strip metadata byte-for-byte without touching pixels, or re-encode and apply a fingerprint perturbation that breaks the original file's hash.

PDF cleaning too

Removes PDF document info (author, creator, producer, dates) and XMP packets while keeping every page intact.

Verified clean output

Cleaned files are re-parsed segment by segment. If anything removable remains, the file is flagged instead of offered for download.

How it works

From inspection to a verified clean file in three steps.

Add your files

Drop a JPEG, PNG, WebP or PDF. The tool immediately inspects it locally and lists every metadata field it finds, grouped by category.

Pick a cleaning mode

Choose lossless stripping to keep pixels identical, or re-encoding to also change the file's fingerprint. PDFs are always cleaned losslessly.

Download the clean file

After cleaning, the output is re-scanned and verified. Download it individually, or as a ZIP for multiple files.

Where it helps

Metadata travels further than most people expect.

📍

Photos with GPS before sharing

🤖

AI images carrying C2PA or prompts

🖼️

Images re-published across platforms

📄

PDFs with author and editor info

🕵️

Journalism, legal and research files

🏠

Property photos with device data

📱

Uploads to social networks

🛒

Marketplace and client deliveries

Formats and limits

What the cleaner accepts and where the boundaries are.

Supported formats

JPEG, PNG and WebP images, plus PDF documents. Other types are rejected before processing.

Size limits

Up to 25 files per queue, 25 MB per file and 150 MB in total. Images larger than 24 MP or 16,384 px are downscaled in re-encode mode.

Single free, batch on a plan

Cleaning one file at a time is free without an account. Queuing multiple files or downloading them as a ZIP requires a subscription.

Local by design

Because processing happens in the tab, there is no upload progress, no server storage and no retention period to trust.

What it can't do

Honest limits — this tool removes container metadata only.

Invisible pixel-level watermarks embedded in the image itself — such as SynthID — are not metadata and are not removed or modified.

AI detectors that judge the image content (not its metadata) can still flag a cleaned image. This tool removes metadata, not detection signals in pixels.

In PDFs, the document info and XMP packet are removed, but images embedded inside pages are not rewritten and keep their own data.

Lossless mode keeps compressed pixels identical. Re-encode mode necessarily re-compresses the image — quality stays visually high, but the file is not bit-identical.

Frequently asked questions

What the cleaner does — and doesn't do.

Clean a file in seconds

Drop a file above — inspection and cleaning happen locally, and the result is verified before you download it.

Clean metadata now

Popular AI Tools

Try other online video and image tools that work in your browser.