AI Detection Crisis: Tools Failing, Distrust Persists

New analysis argues detection is dying due to high misclassification rates (61% for non-native speakers) and OIA rulings, not pedagogical shift. Illingworth finds policies use educational language for surveillance. A new mixed-methods study offers concrete strategies for bridging AI ethical gaps in authentic assessment, supporting the redesign approach. The sector faces a choice between genuine assessment redesign or replacing one surveillance system with another.