A comprehensive glossary of the visual glitches, logical errors, and tell-tale signs found in AI-generated imagery. Master these concepts to become an expert spotter.
When shadows or light sources in an image do not align logically, implying multiple suns or impossible physics.
Background elements that are blurry, warped, or architecturally impossible.
Reflections in mirrors, water, or eyes that do not match the scene.
The failure to maintain repeating details like windows, brickwork, or fabric prints across a surface.
Different materials (e.g., skin and clothes) merging into a single surface.
Earrings, glasses, or clothing details that do not match on the left and right sides.
Invisible grid-like patterns detectable only with Fourier analysis tools.
The tendency of AI to default to specific "pleasing" aesthetic styles (orange/teal, dramatic lighting).
Eyes that clearly differ in shape, size, pupil structure, or iris color.
Hands with too many digits, missing knuckles, or impossible joints.
Hair strands that defy gravity, turn into other objects, or lack a root source.
Skin texture that lacks pores, fine lines, or natural imperfections, appearing like wax or plastic.
An overload of teeth, or teeth that lack size consistency.
Background text that resembles real letters but forms no coherent words.
When the visual features of one concept "bleed" into another due to semantic proximity in the model.
The absence of EXIF data (camera model, shutter speed) in digital files.
The new industry standard for cryptographically proving an image's origin.
A verification technique to find the original source or early instances of an image.
Invisible noise patterns embedded into AI images to identify them as synthetic.