Thus far these detectors kind of suck, both for deepfakes and AI generated text. They’re biased against non-native speakers and using them in a scholarly setting can result in punishing students that aren’t cheating.
The genie was let out of the bottle much too early.