Waterloo Ends Turnitin AI Detection

Waterloo’s decision matters because AI detection is sold as an easy add on to existing plagiarism workflows, but schools can remove it quickly if faculty stop trusting the signal. Turnitin’s core text matching product stays embedded in campus systems, but the higher value AI layer is more fragile because it depends on instructors believing the output is fair enough to act on in real student cases.

• Waterloo did not just pause the feature quietly. It said the tool would be discontinued in September 2025 after consultation across academic leadership, citing unreliability, bias concerns for non native English speakers, inconclusive internal testing, and cases where human writing was flagged as fully AI generated.
  1 uwaterloo 2 uwaterloo
• That is a real monetization risk because Turnitin has already shown schools will pay extra for AI detection. California State University paid an additional $163,000 in 2025 for the AI add on, on top of its broader Turnitin contract. If more institutions follow Waterloo, add on revenue weakens even if the base plagiarism product remains sticky.
  3 sacra
• The opening this creates is mostly for workflow shifts, not instant vendor replacement. GPTZero and similar specialists compete on more transparent sentence level signals and authorship logs, while many schools are moving toward draft history, assignment redesign, and AI disclosure rules instead of trying to prove misconduct from one detector score.
  4 sacra 5 sacra 1 uwaterloo

The next phase of this market is likely to center less on catching AI after submission and more on proving how work was produced. That favors products that record writing activity, show draft history, and fit new academic integrity policies, while pushing pure detection tools, including Turnitin’s add on, into a narrower supporting role.