Metadata investigation
Every file is hiding something.
Drop an image to see the metadata and AI provenance it carries, or paste text to find the invisible characters hiding in it. Both run entirely in your browser.
Ready. Choose an image to scan.
🔒 Your file stays on your device.
JPG, PNG and WebP · up to 25.0 MB · nothing is uploaded
🔒 Processed locally in your browser. Your text is never sent anywhere.
This does not remove statistical text watermarks.
A statistical watermark is encoded in a model’s word choices, not in invisible characters. Removing hidden Unicode has no effect on it, and no browser-side tool can confirm whether one is present. NoWatermark does not claim to detect or remove Anthropic’s statistical watermark.
What NoWatermark checks
AI provenance
C2PA manifests, Content Credentials and the generator tags AI tools leave behind — including the prompt and settings stored inside PNG files.
Private metadata
GPS coordinates, camera and phone identifiers, timestamps, editing software and authorship fields — the data a photo carries without showing you.
Hidden characters
Invisible Unicode, zero-width spaces and bidirectional controls that survive copy and paste, in both file metadata and pasted text.
Popular tools
All toolsAI Watermark Checker
See every provenance and metadata signal in an image.
AI Watermark Remover
Strip AI provenance metadata without touching image quality.
ChatGPT Checker
Look for OpenAI provenance data in an image.
Claude Checker
Find invisible characters hiding in text.
C2PA Checker
Detect and read Content Credentials manifests.
AI Metadata Remover
Remove prompts, model names and generation settings.
EXIF Remover
Strip GPS, device and timestamp data from photos.
Your files stay yours
NoWatermark has no upload endpoint, no database and no file storage. Your image is read from disk into your browser's memory, inspected there, cleaned there, and offered back to you as a local download. Nothing is transmitted at any point.
You do not have to take our word for it. Load this page, disconnect from the internet, then scan and clean a file. Everything works, because it was always working locally.
🔒 Processed locally in your browser
Not every AI watermark works the same way
Metadata can be removed
EXIF, XMP, IPTC, PNG text chunks and C2PA manifests are records attached alongside the picture. They can be read, removed cleanly, and — crucially — the removal can be verified by scanning the result again. That is what our cleaners do.
Embedded signals cannot
SynthID and similar techniques encode a signal into the pixels themselves. Metadata cleaning never touches pixels, so it cannot affect them — and no browser tool can even detect them. We report those as Unable to verify, never as absent.
Popular guides
All guidesWhat is C2PA?
The provenance standard behind Content Credentials, how it is stored inside an image, and the two places it tends to break down.
What is SynthID?
Google's imperceptible watermark lives in the pixels, not the metadata — which changes everything about what you can do with it.
Does Claude watermark text?
Two very different things get called "AI text watermarks". Only one of them is something you can find in your browser.
Does ChatGPT watermark images?
ChatGPT images have generally carried C2PA Content Credentials — metadata, not a visible mark — and metadata is easily lost.
C2PA vs SynthID
One is a label on the bottle, the other is dissolved in the liquid. Almost every practical question follows from that distinction.
Can you remove AI watermarks?
Some can, some cannot, and the difference is entirely predictable once you know where the signal lives.
Frequently asked questions
Are my files uploaded?
No. Every scan and clean runs inside your browser. There is no upload endpoint and no storage. You can disconnect from the internet after the page loads and the tools keep working.
Can you remove SynthID?
No. SynthID is embedded in the pixels rather than the metadata, and no browser-based tool can detect it — which also means none can confirm removing it. We report it as "Unable to verify" on every image.
Does cleaning reduce image quality?
No. Metadata is removed at the container level and the compressed image data is copied byte-for-byte, so the pixels are identical afterwards.
What does the text checker find?
Invisible Unicode - zero-width spaces, bidirectional controls and Unicode tag characters that survive copy and paste. It deliberately preserves the zero-width joiners and variation selectors that emoji need, so cleaning does not break them. It does not detect statistical AI-text watermarks, and no browser-based tool can.
Which formats are supported?
JPG, PNG and WebP for images, and any pasted text for the hidden-character tools. More formats may follow.
Do I need an account?
No. Every tool works without registration, and nothing is stored.