See exactly what is inside your file
Before you remove anything, find out whether there is anything to remove. The inspector parses your image or video in this tab and names every provenance marker it carries, with its size and, where the payload says so, the tool that wrote it.
Nothing is uploaded
Every byte is parsed in this tab. There is no server to send a file to, so there is nothing to trust us with.
Lossless by construction
JPEG, PNG, WebP and MP4 are rewritten at the container level. The image and video data is copied byte for byte.
You see what was found
Every file is inspected before and after, and each finding is named with its size. No silent operations.
No account, no limit
Fifty files at a time, any size your browser can hold, free. No sign-up wall in front of the tool.
Every marker a platform reads at upload
Each of these is detected, named and reported with its size before it is taken out -- and each can be switched off if you want to keep it.
C2PA content credentials
Signed provenance describing the tool, the account and the edit history. The single strongest label trigger.
XMP packet
Adobe's metadata format. Carries generator names, model versions, prompts and edit descriptions.
EXIF metadata
Camera and software fields. Also where AI editors leave their signature, alongside your GPS coordinates.
IPTC attribution
Includes digitalSourceType, the field that says an image came from a trained algorithm.
Text chunks
PNG tEXt/iTXt and JPEG comments. Where Stable Diffusion and ComfyUI store prompts, seeds and whole workflows.
Container metadata
MP4 udta and meta boxes: editor tags, device info, project traces.
Four layers sit between your picture and the label
A tool that treats all of them as metadata is selling false confidence. Here is exactly which two we address, and which two nobody can address from a browser.
Visible pixels
What a person sees. Crops, filters, burned-in logos and subtitles. Changing them proves nothing about the hidden layers.
Pixel watermarks
SynthID and similar systems embed a signal into the image data itself, designed to survive editing. No browser tool removes these, including this one.
C2PA content credentials
A cryptographically signed record of origin and edit history. This is the strongest single trigger for an automatic label.
EXIF, XMP, IPTC and text
Ordinary file metadata: camera fields, GPS, editor names, prompts, seeds, workflow dumps and the IPTC field that marks an image as algorithmic.
Frequently asked questions
What is the difference between the checker and the cleaner?
The checker only reads. It parses the file, lists what it finds and modifies nothing. The cleaner uses the same parsers and produces a new file with the findings removed. Running the checker first tells you whether cleaning is even needed.
Does the checker change my file?
No. It never produces an output file and never writes to disk. It reads bytes into memory in this tab and reports what they are.
It says my file is clear but the label still appeared. Why?
Then the trigger was not file metadata. The remaining causes are a pixel watermark such as SynthID, a visual classifier, your own disclosure toggle, or the caption text. None of those live in the container, and no metadata tool can address them.
Can it tell me which AI tool made an image?
Often, yes - when the payload names itself. We surface that name and whether prompt data is present, read from strings already inside your own file. What we do not do is dump the full metadata, because a signal report is what is useful here.
Which formats can it inspect?
JPEG, PNG, WebP, AVIF, HEIC, MP4 and MOV. Format detection is done from the file's magic bytes, so a mislabelled extension does not fool it.
Found something? Clean it.
The cleaner uses the same parsers, so what the checker listed is exactly what comes out - no surprises between the two screens.