I van totally believe that it detects AI generated content 99% of the time, that’s trivial. What I really wanna know is the false positive rate. If I write a program that flags everything, it’d have a 100% hit rate. It’d also however have a crazy high false positive rate.