日本語

Examine an image

Four independent layers of evidence about one image, each explained. It tells you what the file contains and where recompression looks uneven. It does not tell you whether the image is fake, because no honest tool can.

Loading the tool…

How it works

  1. Drop the image. The four layers run in order.
  2. Read the ELA map with its guide — bright areas are places to look, not proof.
  3. Check the metadata, the Content Credentials chain and any generator markers, then read the summary.

Why nothing is uploaded

Every operation on this page is done by code running inside your browser tab, using the same engine that renders web pages. The file is read from disk into your tab’s memory, transformed there, and written back out as a download. It is never sent anywhere — not to us, not to a third party.

Verify it yourself

  1. Open your browser’s developer tools (F12) and select the Network tab.
  2. Load your file and run the tool.
  3. The only requests you will see are the tool’s own code and, once cached, none at all.

Proof it stays local →

Frequently asked questions

Why is there no “is this AI?” percentage?

Because we could not find a model worth shipping. The credible detectors are licensed for non-commercial use only; the permissively-licensed ones have a handful of downloads and no published evaluation. A confident number from a bad classifier is worse than silence on a question people act on. What is shown instead are facts: signatures, embedded generator markers and recompression evidence.

What does error level analysis actually show?

The image is recompressed at a known quality and compared with itself. Areas that survive the second compression differently from their surroundings light up. A region pasted in from another JPEG usually differs — but so do sharp edges, text and fine texture. It shows where to look, never what happened.

The image has no metadata. Is that suspicious?

Almost never. Every major social network and messaging app strips metadata on upload, so the absence tells you the image passed through one of them, and nothing more. This is stated on the page because “no EXIF” is the most commonly over-interpreted finding in amateur image forensics.

What are Content Credentials?

C2PA: a cryptographically signed record of how an image was produced and edited, attached to the file. When present it is strong evidence, because the signature can be verified and the chain of edits read. Very few images carry it yet, so its absence means nothing.