Turnitin AI Detection Flags Wrongly Accuse Honest Students
Turnitin's AI-detection service wrongly flagged a student's honest work as AI-generated, resulting in a zero grade. Research shows AI-detection services have a 6.8% false positive rate, causing widespread anxiety among students who fear academic consequences for work they did themselves.
What happened
Leigh Burrell, a computer science major at University of Houston-Downtown, received a zero on a mock cover letter assignment worth 15% of her final grade in a required writing course after being flagged by Turnitin's AI detection service.
According to Google Docs editing history reviewed by The New York Times, Burrell had drafted and revised the assignment over the course of two days, not using AI. She appealed the decision and her grade was restored after providing a 15-page PDF of time-stamped screenshots and notes documenting her writing process.
A study of a dozen AI-detection services found that they had erroneously flagged human-written text as AI-generated 6.8% of the time, on average.
Reported impact
Students reported persistent anxiety about being accused of using AI on work they had completed themselves, and facing potentially devastating academic consequences.
Many students have imposed methods of self-surveillance they describe as self-preservation, such as recording their screens for hours while doing schoolwork or composing papers using word processors that track keystrokes closely.
Index analysis
The incident reflects broader tensions between AI cheating concerns in education and the limitations of detection systems. According to a Pew Research survey, 26% of teenagers said they had used ChatGPT for schoolwork, double the rate of the previous year, prompting educators to adopt detection tools.
The false positive rate of AI-detection services and the burden placed on students to disprove accusations suggests that imperfect systems may affect students who are following academic rules.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.