What WSU did
On February 13, 2026, WSU's Faculty Senate published notice of the cancellation, and the Provost's office updated its policy on detecting and reporting AI-related misconduct. The student newspaper, the Daily Evergreen, covered the decision with the headline "Turnitin is out."
The number that did it: 33%
WSU's own published reasoning contains the decisive statistic. Between 2023 and 2025, 33% of all Review Board cases alleging AI misuse ended in a "not responsible" finding, because the only evidence offered was a Turnitin AI score. One accused student in three, cleared, after going through a formal misconduct process that the detector alone had triggered.
Think about what that 33% actually measures. It isn't the detector's error rate. It is the rate at which the institution's own process, when it finally looked closely, refused to stand behind the detector. Faculty were using the score as sole evidence even though Turnitin's own documentation says not to. The tool invited a workflow its own vendor disclaims.
The Daily Evergreen ran a second number: WSU processed 148,547 assessments through Turnitin in Fall 2024, and applying Turnitin's own published document-level false-positive rate of just under 1% yields roughly 1,485 essays likely false-flagged in a single semester. Treat that figure carefully. It is the student paper's extrapolation, not an official WSU count, and Turnitin's published rate technically applies only to documents scored at 20% or more AI. But even read skeptically, the order of magnitude is the point: at university scale, a "small" error rate is a four-digit number of students per term.
What replaced it
WSU didn't replace one detector with another. The provost's guidance moved to SafeAssign for traditional plagiarism and, for AI concerns, to faculty-led review of the circumstances: drafts, process artifacts, the student's account, and conversation. In other words, the university concluded that judging authorship is evidentiary work, not something to outsource to a probability score.
What K-12 teachers should take from this
WSU had advantages a K-12 district doesn't: a formal review board, due-process machinery, and the institutional capacity to absorb a 33% failure rate while studying it. A high-school student facing a flag usually gets a conversation with one teacher and a grade decision the same week. The same failure mode, with fewer safeguards, is already producing K-12 casualties: a federal lawsuit in Palo Alto, a school-board petition in Wake County after three detectors gave the same essay three contradictory scores, and state guidance (North Carolina's, among others) now warning that detectors are not dependable.
The takeaway is not "use detectors more carefully." WSU tried that; the December 2023 vendor warning was already on the books while the 33% accumulated. The takeaway is that a score is the wrong category of evidence, and the institutions with the most experience are now saying so out loud.
The K-12-friendly version of what WSU moved to
WSU's replacement, faculty-led process review, assumes the process evidence exists. For a university with drafts and submission systems, sometimes it does. For a classroom teacher, it can simply be built in: assign writing in an environment that records the process as it happens, and every submission arrives with its own evidence attached. No review board needed; the replay answers the question a hearing would have litigated.
That is what Manupropria does, at single-teacher scale and a single-teacher price. If your district is having its own version of the WSU conversation, the wider landscape of alternatives is here.