Study reveals GPT detectors misclassify non-native English writers’ work as AI-generated, highlighting significant bias.