ChatGPT Detectors Are Biased and Easy to Fool, Research Shows - CNET

Stanford researchers find GPT detection software routinely misclassifies writing from non-native English speakers and can be duped by "literary language."