When Turnitin rolled out AI detection in 2023, professors began failing and reporting students for AI writing. But the tool had a high false positive rate: real students with crisp, structured writing styles were being flagged as AI. A UC Davis professor publicly defended a student whose original thesis was flagged as 'likely AI.' Students with dyslexia who relied on writing tools, non-native English speakers whose grammar was unusually correct, and prolific writers who had developed clean prose styles were all disproportionately targeted. Turnitin acknowledged the issue but maintained the tool was accurate 'at scale.'