Each one uses the same principle as everything else here: show the evidence, explain the reasoning, and be honest about the limits. No black-box verdicts.
Paste in any piece of writing and see which patterns of AI-generated text it contains. Instead of a single percentage, you get every signal it found, explained, so you can judge for yourself. Built on the methodology Wikipedia's editors documented.
Upload an image and we read its metadata for traces of AI generators, including C2PA Content Credentials, EXIF signatures, and embedded Stable Diffusion prompts. Honest about its limits: strong positive signals, no false certainty.