AI detector

AI text detectors score writing on how predictable it looks, on the theory that AI writing is smoother and more uniform than people's. The trouble is that plenty of human writing is smooth and uniform too: careful academic prose, formulaic reports, and especially writing by people working in a second language. Meanwhile, light editing can carry AI text past them.

So text detectors raise false alarms and miss real cases, and a score can't settle whether someone used AI. For images, video and audio, the more dependable checks are marks added when the content was made, such as Content Credentials and watermarks, and those only work on content that carries them.