UCT Scraps Turnitin's AI-Detection Score — A South African University Bets on Assessment Redesign Over Surveillance

In June 2025, UCT's Senate Teaching and Learning Committee endorsed a new AI in Education Framework; from 1 October 2025 the university stopped using AI-detection tools, including Turnitin's AI Score, in student assessment, citing unreliable results (both false positives and false negatives) that risked wrongful misconduct accusations, particularly against multilingual and non-native-English writers.

Replace automated AI-detection surveillance with an approach built on ethical AI literacy, transparency about permitted AI use, and assessment redesign that is harder to complete dishonestly with AI alone.

UCT withdrew Turnitin's AI Score and other detectors from assessment processes and rolled out its AI in Education Framework university-wide, against a national backdrop where peer universities take different approaches: the University of Pretoria recorded roughly 53 AI-related disciplinary cases in 2024-2025 and triangulates evidence (declarations, version history, oral clarification) rather than relying on a single detector score, the University of Johannesburg treats undisclosed AI use as fraudulent authorship, and Unisa combines AI-enabled proctoring with tiered disciplinary responses.

The decision was widely covered in South African media as a landmark break from a widely used but increasingly distrusted detection product; no measured learning-outcome data has been published.

UCT's move illustrates South African universities independently converging on the conclusion reached by regulators elsewhere - that detection software alone cannot reliably police AI use in assessment.