Universities that turned off AI detection: a verified list

The short answer

At least nineteen universities — Vanderbilt, Yale, Georgetown, NYU, Johns Hopkins, UC Berkeley, and Boston University among them — have switched off, cancelled, declined to license, or refused to enable Turnitin's AI writing detector. Every entry below was verified against the institution's own published page as of November 1, 2026 — not against press coverage or other people's lists — and the quotes are the universities' own words. Third-party trackers that include news-sourced entries count considerably more: one US-only tracker lists 38 institutions, another counts over 60 internationally.

This matters for two audiences. If you have been flagged, the fact that peer institutions reviewed the same tool and shut it off is directly relevant to your response. If you set policy, these are the review outcomes of the institutions that looked hardest.

United States

Institution	What happened	When	In their own words
Vanderbilt	Disabled	Aug 2023	"…disable Turnitin's AI detection tool for the foreseeable future"
Boston University	Disabled	Jul 2023	Disabled "based on a range of concerns, including false positives and lack of supporting evidence for its efficacy"
University of Arizona	Disabled	Jun 2023	Feature "temporarily disabled for all UArizona courses"
University of Pittsburgh	Disabled	2023	"…the Teaching Center has disabled the AI-detection tool in Turnitin"
Georgetown	Disabled	Oct 2023	"…the harms of false positives are worse than any advantages of the tool"
Montclair State	Discontinued by provost memo	Nov 2023	AI detection "officially discontinued in 2023"
CU Boulder	Declined to license after trial	Dec 2023	Advisory group "unanimously recommended not to support" the feature
UIUC	Deactivated campus-wide	—	"Given the AI detection tool's lack of consistency in its results, the campus deactivated this feature"
RIT	Removed	May 2024	"RIT has removed the AI detection feature within Turnitin as of May 13, 2024"
Johns Hopkins	Disabled	—	Disabled "due to reports of false positives and ongoing fears of falsely accusing students"
Yale	Disabled; detection not supported	—	"The AI detection feature of Turnitin is currently disabled"
UC Berkeley	Opted out after a pilot	May 2025	"UC Berkeley has opted out of the use of Turnitin's AI writing detection tool"
NYU	Disabled	—	"AI Writing Detection has been disabled in the Turnitin Similarity Reports"
Washington State	Contract cancelled	Feb 2026	"WSU has cancelled its contract with Turnitin for its AI Detection software"

Canada, UK, and Australia

Institution	What happened	When	In their own words
UBC	Never enabled (decision affirmed)	Apr 2023	"UBC has decided not to enable Turnitin's new AI-detection feature"
University of Manchester	Does not use detection software	Jan 2024	"The University does not currently use software which claims to detect Generative AI"
ANU	Not used for integrity cases; license lapsed	Jan 2024	Tool "will not be used at ANU for the purposes of pursuing academic integrity matters"
University of Waterloo	Discontinued	Sep 2025	Function "no longer be available to University of Waterloo users as of September 2025"
Curtin University	Disabled across all campuses	Jan 2026	"From 1 January 2026, the AI writing detection feature in Turnitin will be disabled"

Where the "When" column is blank, the university's page confirms the decision but does not state the date it took effect; we report only what the official page says.

Read the reasons, not just the list

Lay the universities' statements side by side and the same three findings repeat, in institution after institution, across three years and four countries:

1. False positives land on real students. Vanderbilt published the arithmetic: at the vendor's claimed 1% false-positive rate, its ~75,000 annual submissions would mean roughly 750 wrongly flagged papers per year. Georgetown weighed the same trade and concluded the harms of false positives exceed any benefit. Johns Hopkins cited "ongoing fears of falsely accusing students."
2. The scores aren't consistent. UIUC's stated reason is the tool's "lack of consistency in its results." Boston University cited "lack of supporting evidence for its efficacy." CU Boulder's academic-technology advisory group reviewed a trial and voted unanimously against it.
3. The decisions are sticking and spreading. This started as a 2023 wave (BU, Arizona, Vanderbilt, Georgetown) and has kept going: RIT in 2024, Berkeley and Waterloo in 2025, Curtin and WSU's contract cancellation in 2026. UBC made its decision, took a second look after pushback, and affirmed it.

What this list does not mean

Honesty cuts both ways. None of these universities decided AI misuse is fine, and none stopped enforcing academic integrity. What they stopped doing is treating a detector score as evidence. Integrity cases at these institutions now run on the things that were always stronger: process evidence, drafts, version history, and conversations about the work — the approaches we compare honestly in alternatives to AI detectors.

Also: your institution still using a detector is not malpractice by itself. Even Turnitin's own guidance says the score should not be the sole basis for action. The problem is a score treated as a verdict — which is exactly the pattern a New York court rejected in February 2026 when it ordered Adelphi University to expunge a student's AI-cheating finding.

How to use this list

If you've been flagged: don't wave the list around as "detectors are banned" — they aren't. Use it precisely: institutions including Vanderbilt, Georgetown, and Boston University reviewed this tool and disabled it over false positives and reliability, the vendor itself cautions against sole reliance, so the score needs corroboration your accuser doesn't have — and then present your own process evidence, which is what actually wins cases.

If you set policy: the pattern above is a peer-review verdict rendered by the institutions that studied the tool longest. The follow-up question — what fills the gap — has better answers than "trust the score" or "trust nothing": process-based assessment on the teaching side, and verifiable writing provenance on the evidence side, where students who want protection carry a sealed, independently verifiable record of their own writing process.

We will keep this page updated as official announcements land; corrections with a link to an institution's own page are welcome.