AI art detector
Illustration is harder to assess than photography, because the usual shortcut does not apply: a legitimate digital painting has no camera EXIF either, so its absence means nothing here. What remains is the generator's own metadata and the classifiers.
A genuine digital artwork often carries traces of the tool that made it — Photoshop, Procreate, Clip Studio — in its metadata, and that is treated as meaningful supporting evidence. Generated work frequently carries the opposite.
Drag a file here, or
JPEG, PNG, WebP, AVIF, PDF or .txt · up to 10 MB · you can also paste a screenshot
By uploading you agree to our Terms and Privacy Policy. Your file, IP address, approximate location and device details are stored.
What this checks
- Editing-software tags from Photoshop, Procreate, Krita and similar tools
- Generator names in EXIF, XMP or PNG text chunks
- Embedded generation parameters from diffusion front-ends
- Layer and edit-history traces where the format preserves them
- Image classifiers, weighted for illustration rather than photography
Frequently asked questions
Can it tell AI art from digital painting?
When metadata survives, usually yes — the tool that made it is often named outright. On a bare PNG with no metadata, the classifiers are working at the edge of their reliability and the result will say so.
Why are artists flagged incorrectly?
Highly rendered digital painting shares surface characteristics with diffusion output: smooth gradients, clean edges, even detail. Classifiers key on exactly those features, which is why a metadata-first approach matters for artwork.
Does it work on anime and stylised art?
Less reliably. Most detectors are trained predominantly on photorealistic imagery, so stylised work sits outside their strengths and should be read with extra caution.