AI writing detectors are ineffective, unreliable and harmful

Purpose This paper aims to examine the deployment of artificial intelligence (AI) writing detectors in English language teaching (ELT) contexts and their contradictory impact on pedagogical goals of differentiated, student-centered learning approaches. Design/methodology/approach This paper uses an essay-style approach, drawing on existing literature and documented cases on AI writing detectors. It is organized around three interrelated issues: ineffectiveness, unreliability and harm, with each section defining and exemplifying the concept. Findings AI writing detectors demonstrate significant technical flaws including high false positive rates, bias against non-native English speakers and inability to keep pace with evolving technologies. These tools disproportionately flag authentic writing by multilingual students, creating a chilling effect that paradoxically encourages AI use to avoid false accusations. Reliance on these detectors contradicts essential ELT principles of honoring student voice, promoting linguistic diversity and fostering inclusive learning environments. Originality/value This essay contributes to critical discourse on AI detection in educational contexts by highlighting the inefficacy, unreliability and pedagogical harm caused by surveillance-based assessment approaches. It proposes redefining pedagogical practices in English teaching through introducing critical AI literacy and human-centered assessment practices that prioritize student agency and multilingual expression over technological policing.