AI text detectors used by universities fail at rates too high to justify career-ending misconduct rulings, a new IEEE study found. False positive rates hit 68.6%, and some tools missed nearly all AI-generated text.
Study finds AI text detectors too unreliable for academic use
