Methodology & limitations

The detector looks for statistical/visual patterns often seen in synthetic images. Results depend on image quality and the type of content.

What affects accuracy

• Compression and re-uploads (social apps can alter the image).
• Resolution (very small images contain less signal).
• Content type (portraits vs landscapes vs graphics).
• Editing (filters, sharpening, denoise, upscaling).

False positives / false negatives

A real photo can look AI-generated (false positive) and an AI image can look natural (false negative). Use the tool as one signal among others.

Privacy note

Check the Privacy page for details on how uploads are processed and whether any data is stored.