Preprint
GPT detectors are biased against non-native English writers
Weixin Liang et al., arXiv (later published in Patterns), 2023
Read the originalRead on October 17, 2026
- What it is
- A Stanford study that tested several widely used GPT detectors on writing samples from native and non-native English writers. The authors report that the detectors consistently labelled the non-native samples as AI-generated.
- How much weight it carries
- A preprint whose abstract the team later published in Patterns, so the abstract used here is the same finding. The sample was a small set of essays, and the study tested general GPT detectors rather than Turnitin or any dictation tool.
Facts taken from this source
- In a 2023 study of several widely used GPT detectors, the detectors consistently misclassified non-native English writing samples as AI-generated.
“Our findings reveal that these detectors consistently misclassify non-native English writing samples as AI-generated” (Abstract)