Detecting AI content

A claim that keeps failing.

Unreliable

Detection tools produce both false positives and false negatives, and have wrongly accused students, disproportionately those writing in a second language.

Style is not evidence; plain writing is not proof of anything.
Institutions relying on detectors have had to reverse decisions.

Provenance instead

Signed content credentials attached at creation are more promising than after-the-fact detection, and they only work where they are applied.