AI Image Detector
Analyze the pixels themselves - spectral fingerprints, smoothness, and color statistics - plus C2PA provenance metadata. Read in your browser, never uploaded.
Visual analysis
Provenance metadata
The analysis uses statistical signals from image-forensics research - it is not a trained detection model. Every measured value is shown so you can judge the evidence yourself, and no single signal is proof in either direction. Compressed, edited, or stylized images can fool any heuristic.
Two evidence channels
1 - Pixel analysis
The frequency spectrum, smoothness, colorfulness, saturation, and dimensions are measured directly from the image.
2 - Provenance metadata
C2PA records, XMP and EXIF tags, and known generator names are searched inside the file itself.
3 - Honest limits
Every measurement is shown with the lean it supports - statistical signals, not a black-box score.
Image detector questions
How does the AI image detector analyze an image?
It runs two channels in your browser. Pixel analysis measures statistics published in image-forensics research: the frequency spectrum (generated images often show flattened spectra or periodic upsampling spikes), smoothness (generated images often lack sensor grain), colorfulness and saturation tendencies, and whether the dimensions match common generator presets. Metadata analysis scans for C2PA provenance records, XMP and EXIF tags, and known AI generator names. Every measured value is shown with the evidence it leans on.
How accurate is it?
These are statistical heuristics, not a trained detection model. They catch common patterns, but AI models keep improving and can mimic photographic statistics, while compression, screenshots and heavy editing can distort signals in either direction. Treat the result as one signal among several - never as proof on its own.
Is my image uploaded anywhere?
No. The file is decoded and analyzed entirely in your browser - it is never sent to a server or saved anywhere.