Why Are UK Universities Dropping Turnitin's AI Detector? The Reasons in Their Own Words

The main reason is accuracy. Of the 33 UK universities that explain why they do not use Turnitin's AI writing indicator or any AI detector, 29 say the tools are unreliable, produce false positives or risk wrongly accusing a student. Next come scores that cannot be checked or used as evidence (15) and concerns about student data, consent and copyright (13). Only two of the data statements are about Turnitin's own indicator. For which universities have said no, and the seven that use the report, see our tracker.

Below is each reason in the universities' own words, with a link for every statement and a date where the page gives one, followed by the reasons given by the universities that do use the AI report.

The short answer

• Are they really dropping it? Mostly they never switched it on. Turnitin announced it would turn the feature on for all customers in April 2023 unless they opted out, and most statements that name the indicator describe opting out, not dropping it later. For which universities have said no, see the tracker.
• Why? Accuracy first (29 of 33). Then scores that cannot be checked or used as evidence (15), student data, consent and copyright (13), detectors missing AI-written text (9), bias against some writers (9), and a rushed launch and student anxiety (2).
• Is student data the reason? For some. Birmingham and Ulster name privacy or data security about Turnitin's indicator itself. Eleven others raise data protection, consent, copyright or training data about AI detectors in general or about uploading students' work to outside tools.
• Is this about Turnitin's new licence agreement? That is a separate, newer question; see below.
• Why do some still use it? Fairness and extra information for staff, though most of them say a score is not proof. See their reasons.
• Does a score prove misconduct? Not on its own. Fifteen of the universities that say no make the point that a score cannot be checked or used as evidence, and five of the seven that use the report say it is not proof on its own or must be checked by staff. See AI detector scores as evidence in UK universities.

What universities have actually turned down

Turnitin added an AI detection feature in April 2023, as a 2025 report from the University of Edinburgh records. An announcement that month from Bournemouth University says Turnitin "announced it would be used from early April for all Turnitin customers unless they opted out". That is why most universities say "opt out", not "ban", and most statements that name the indicator say it was never enabled. A few universities switched it off after using it: Liverpool piloted it from September 2023 and removed it from 1 August 2024, and Aberystwyth turned it off from 30 September 2023, as its Digital Education update records. Others say they do not use AI detectors at all (Sheffield, UCL) or do not endorse them (Oxford).

These decisions are about the AI indicator or AI detectors, not Turnitin's text-matching. Several of the same pages describe Turnitin's similarity checking as still in use; Dundee's student Turnitin guide calls Turnitin "the main tool that we use for you to submit text-based assignments".

Genutext's UK university AI-detection policy tracker lists who has decided what. This guide answers the other question: why.

How we counted. The tracker records 37 UK institutions that have said no in public (22 with an AI-writing indicator explicitly switched off, and 15 more stating that AI-detection tools generally are not used, permitted, relied on or recommended) and 7 that use an AI writing report, as of 28 August 2026. For each of the 37, we opened the page the tracker cites (or its current location) on 17 October 2026, rechecked several on 18 October, and noted every reason given. Five pages did not open (Glasgow's and Leeds's are gone, Warwick's and City St George's now ask for a login, and Nottingham's is no longer publicly accessible), so for those we used the tracker's August record. Four give no reason (Bath, which uses Inspera Originality rather than Turnitin and does not have its AI authorship detection enabled; Strathclyde; UCL; and Edinburgh), so the counts are out of 33. Imperial explains its choice only by what it does instead. Most give several reasons, so the counts overlap.

Reason 1: false positives and unreliable results

29 of 33 universities. No other reason comes close. Eighteen of them name false positives, or the risk of wrongly accusing a student, explicitly. The rest say the tools are unreliable or inaccurate.

• King's College London (AI guidance, October 2023, updated April 2025): "King's took the decision not to enable the AI detection % in Turnitin due to concerns about its reliability and potential for false positives."
• University of St Andrews (student FAQs, January 2025): "Whilst Turnitin has developed AI-detection, this has not currently been switched on at the University of St Andrews due to concerns over accurate detection."
• University of Reading (academic integrity guidance, November 2023): "Turnitin's facility to detect AI use is turned off due its low reliability and the risk of false accusations of misconduct".

Liverpool, which piloted the tool from September 2023, says its pilot "highlighted a lower-than-expected level of reliability". A false positive means a student who wrote their own work gets flagged; our guide to false positives in human writing explains how it happens, and How accurate is Turnitin's AI detection? sets Turnitin's figures beside independent evidence.

To see which sentences of your own draft read as AI-like before you submit, check it below. Treat the result as a reason to reread a passage, not a verdict; different detectors give different results for the same text.

Reason 2: nobody can check the score, so it is not evidence

15 of 33 universities. A plagiarism match points to a source that a marker can open. An AI score does not. Eight universities say the result cannot be verified, explained or tested before use (Bournemouth, Dundee, Durham, Greenwich, Oxford, Ulster, USW and Warwick). Eight say a score cannot serve as evidence of misconduct (Bath Spa, Cambridge, Manchester, Middlesex, Nottingham, Oxford, Sussex and UAL). Oxford says both.

• Durham University (DCAD FAQs, no date on page), asked whether it has the Turnitin AI detector: "No we don't." It explains: "Unlike accusations of plagiarism where the source documents exist, you can't definitively match student work back to generative AI output. The detection tools are closed systems which makes them exceptionally hard to verify."
• University of Dundee (AI in teaching and assessment, updated 16 September 2026): "A key factor in this decision was that we were unable to test and evaluate the reliability of the detection tool and were not provided with independently verified data to give us sufficient confidence to implement at this time."
• University of Cambridge (FAQ, no date on page): "Detection tools offer no evidence to support investigations even when identifying possible use in a submitted example."

Reason 3: student data, consent and copyright

13 of 33 universities. If you searched for universities dropping Turnitin over student data, this is the part of the AI-detection story that is about data. The licence-agreement story is separate and covered below. The statements fall into three groups.

About Turnitin's own indicator: 2.

• University of Birmingham (generative AI guidance, July 2023, updated January 2025) says it has not enabled Turnitin's AI writing detection, and: "There remains a need to better understand its effectiveness and to assess privacy and data security considerations arising from its use."
• Ulster University (staff guidance, undated): "The AI working group, at Ulster, had many concerns from an ethical, accuracy and privacy perspective and made the decision to disable this tool."

About AI detectors in general: 4.

• University of the Arts London (UAL) (AI and teaching, undated) gives three reasons for using no AI detection software. Two are about data: "Uploading the material to these platforms could lead to a breach of data security and copyright" and "The content submitted can be used as training data for other AI systems."
• Edinburgh Napier University (position on generative AI, August 2025) does not permit detection tools on student submissions. Part of its reason: "we do not have students' informed consent for their use with their work".
• Middlesex University (AI policy, published September 2026) says there are "significant concerns about the GDPR compliance of AI detection tools". The fourth is Leeds, whose educator guidance cited data security and training-data exposure when the tracker recorded it in August; that page no longer loads.

About sending students' work to outside tools: 7. Rules for staff at Portsmouth, Reading, Manchester Met, Brunel, Sussex, Dundee and City St George's (for doctoral research) say students' work must not go to outside detectors or unlicensed AI tools. Their reasons are data protection (Sussex), data protection and integrity (Portsmouth), IP and GDPR (Brunel), insecure data (Manchester Met), no student consent (Dundee), confidentiality and student IP (City St George's, as the tracker recorded it), and uncertain legal and ethical implications (Reading).

• University of Sussex (staff guidance, undated): uploading student submissions to AI tools it does not license "contravenes the University's data protection rules".
• Manchester Metropolitan University (AI in education, undated) says of online AI detection services: "These tools are not reliable and the data uploaded to them is not secure."
• University of Dundee (teaching guidance, updated 16 September 2026) tells lecturers not to use unauthorised AI detection tools: "We have not approved any of these tools and we do not have student consent to upload their work to third party sites."
• University of Portsmouth (staff AI guidance, no date on page): "AI tools are not currently permitted for uploading student work, marking or feedback due to data protection and integrity concerns."

Others, such as Glasgow and Strathclyde, have similar no-upload rules but give no reason for them.

None of these statements says Turnitin has misused student work. Data protection is a real reason, but far fewer universities cite it than cite accuracy, and most of the concerns are about where staff send a student's work rather than about the Turnitin indicator.

Reason 4: it misses AI-written text too

9 of 33 universities. A detector can be wrong in both directions. Nine universities point to false negatives, or say detection cannot keep up with AI tools (Bath Spa, Birmingham, Brunel, Durham, Liverpool, Oxford, Reading, Sheffield and USW). Brunel turns this into a fairness point: "AI detectors are relatively easy to fool, and therefore only weaker students will be caught and punished." The University of South Wales (USW) says in its AI detectors statement (early 2024) that detectors are "unable to keep up with the rapid development of AI".

Reason 5: bias against some writers

9 of 33 universities. Four say detectors are biased against students whose first language is not English (Liverpool, Manchester Met, Sussex and USW). Oxford's AI Competency Centre article has a section headed "Bias Against Vulnerable Students", naming students who speak English as an additional language or are neurodivergent. Goldsmiths names students who used AI translation tools. Durham cites bias against certain demographics. Manchester and Edinburgh Napier call the tools biased.

• Manchester Metropolitan University (AI in education, undated): "The University chose not to enable the AI detector launched by Turnitin in 2023 because of its rate of false positives, and because of evidence that such tools are biased against writers whose first language is not English."
• University of South Wales (AI detectors statement, early 2024) says detectors "Frequently flag more simple language as being AI generated when it is not", adding: "Not only does this represent a bias against students whose first language is not English".
• Goldsmiths, University of London (AI in assessment library guide, last updated 23 September 2026): "Goldsmiths and other universities have not enabled it mostly due to accuracy concerns and unfairness towards those who have used AI-powered translation tools in their submissions."

Turnitin has disputed this. Durham's FAQ notes that Turnitin claimed in 2023 that its detector is not affected, and Northumbria's student guide says Turnitin "have taken into account under-represented groups like second language learners". Our guide to AI detectors and non-native English writers covers the research.

Reason 6: a rushed launch and student anxiety

2 of 33 universities, both writing in 2023.

• Bournemouth University (13 April 2023): "We have decided that introducing the functionality at this time would not be fair or appropriate for either staff or students." It adds that Turnitin introduced it "on very short notice, with limited information and without the opportunity for preview or testing by individual universities".
• Aberystwyth University (27 September 2023) turned the tool off, citing "statistics that seem to demonstrate a high incidence of false positives and the anxiety that this induces for students."

What they do instead

Saying no to a detector does not mean saying no to checking work. Seven universities tie their decision to assessment design, or to judging what the student has actually learned (Bath Spa, Cambridge, Durham, Imperial, Middlesex, Oxford and Reading).

• Bath Spa University (AI in teaching, no date on page): "Considering the design of assessment will be more important than trying to retrospectively identify the use of GenAI."
• University of Oxford (AI Competency Centre, 16 February 2026): "The solution isn't better detection. It's better pedagogical and assessment design."
• Imperial College London (published FOI information, entry dated September 2026): "We focus on evidence of student learning rather than automated detection." Its list includes assessment design, review by academic staff and, where appropriate, authenticity interviews or oral follow-up discussions.
• UCL (GenAI and academic integrity, first published 8 October 2024): "UCL does not use GenAI detectors when marking. Instead, if a marker has any concerns about your work, they'll usually request a quick chat to get more information."

Not using a detector routinely is not always the same as never using one. Newcastle's library FAQ says the university "will only employ an AI detection tool as part of an academic misconduct investigation if required".

The universities that use the AI report, and why

Seven institutions in the tracker confirm an AI writing report is in use, by their latest statements: Bristol, Northumbria, Leeds Beckett, Birkbeck and Birmingham City; Abertay, where staff "may consult the AI Writing Indicator but are not required to review it routinely"; and UEA, for plagiarism officers only. Birmingham City's entry rests on the tracker's August 2026 record: the page it cited has since changed, and the quoted sentence is no longer there. The reasons they give are fairness (Leeds Beckett) and more information for staff (Northumbria, Abertay), and five of them (Abertay, Birkbeck, Bristol, Northumbria and UEA) say in the same breath that the score is not proof on its own or must be checked by staff.

• Leeds Beckett University (principles for generative AI, 13 February 2024): "To make sure that nobody gains an unfair advantage from generative AI, we have also introduced the Turnitin AI detection tool in the Academic Year 2023/24 so that any unacknowledged use of this learning resource can be detected."
• Northumbria University (student guide, March 2024): the detector "does not decide if academic misconduct has occurred, rather it provides data for your tutors so they can make an informed decision".
• Abertay University (student guide, undated): the indicator's results "do not prove that AI was used or that academic misconduct occurred and must not be used as the sole evidential basis for a misconduct decision."
• University of East Anglia (generative AI policy, 6 November 2023) approved the software "but this must be used with caution by Plagiarism Officers only due to reported inaccuracy".

Bristol's November 2025 update says results should be "used as supporting evidence, not as a replacement for academic judgement", and Birkbeck's 2025–26 student guidance has staff check a flag "against the permitted use outlined in the assessment brief as well as your statement in the AI Declaration". So most of the universities that use the report treat a score as a prompt for a person to look, not a finding.

Is this about Turnitin's new licence agreement?

That is a separate question. The decisions counted above are about AI detection. By October 2026 a different concern had arisen, about a proposed new Turnitin End-User Licence Agreement (EULA): the agreement that, according to Dundee's student guide, students are shown when they first use Turnitin.

On 5 October 2026, Aberystwyth University's Digital Education team posted Turnitin's response. It says Turnitin "postponed the issue of a new End-User Licence Agreement (EULA) until September 2027" after concerns from UK universities and colleges were "channelled through UCISA". The post also relays that Turnitin "has confirmed that no student or staff data has been used to develop any generative artificial intelligence (AI) models or large language models (LLM), and that they have no plans to use student essays in this way." Those are Turnitin's statements as Aberystwyth reports them, not findings of ours.

On the current agreement, Dundee's student guide says "Point 5 notes that you retain ownership of your work at all times". It adds that some students worry that uploading work gives Turnitin ownership of it. If your university has announced a change to its Turnitin contract, its own announcement is the source to read. None of the statements counted in this guide is about that.

What this means for you

1. Find your own university's position. Look it up in the tracker. Then check your module handbook or assessment brief, which may set its own rules for that assessment.
2. Ask if it is not clear. "Is Turnitin's AI indicator switched on for this module?" is a fair question for a module leader. You may not see the score yourself: Abertay, which uses the indicator, says students "cannot view the AI Writing Indicator or report directly in Turnitin, including after submission". Does Turnitin detect AI? explains what the report checks and who sees it.
3. A score alone is not evidence of misconduct. Most of the universities that use the report say so, and many that do not use it say a score cannot be checked or used as evidence. If your work is questioned, the process matters more than the number, and dated drafts and version history are stronger evidence than any detector. See AI detector scores as evidence in UK universities, our guide for students accused of using AI and how to prove you didn't use AI.
4. If you check your own work, read the tool's data terms. The data rules above are mostly about staff sending a student's work somewhere without consent. Checking your own draft is your choice, but ask the same question: where does the text go? Genutext's privacy policy says submitted text is sent to our analysis provider, Winston AI, is not stored on our servers, and that Genutext does not use it to train or improve AI models. A Genutext score is a signal for your own review and cannot predict what Turnitin will show; our accuracy methodology explains how results are produced.

Frequently asked questions

Why are UK universities dropping Turnitin's AI detector?

Mostly because of accuracy. Of the 33 universities in Genutext's tracker that explain their decision, 29 say AI detectors are unreliable, produce false positives or risk wrongly accusing students. Other reasons are that a score cannot be checked or used as evidence (15), student data and consent (13), missed AI-written text (9) and bias against some writers (9).

Did student data concerns lead UK universities to switch off Turnitin's AI indicator?

Data concerns are one reason, but not the main one. Birmingham and Ulster name privacy or data security as a reason not to use Turnitin's AI indicator. Eleven more raise data protection, consent, copyright or training data about AI detectors in general or about uploading students' work to outside tools. Accuracy is cited far more often.

Did any UK university switch Turnitin's AI detector off after using it?

Yes, at least two. Liverpool piloted it from September 2023 and removed it from Turnitin from 1 August 2024, saying the pilot "highlighted a lower-than-expected level of reliability". Aberystwyth turned it off from 30 September 2023, citing a high incidence of false positives and the anxiety this caused students. Most other statements that name the indicator say it was never switched on.

Why do some UK universities still use Turnitin's AI report?

Those that explain it give two reasons: fairness (Leeds Beckett says it introduced the tool so that unacknowledged AI use "can be detected") and more information for staff (Northumbria and Abertay). Five of the seven in Genutext's tracker that use it also say the score is not proof on its own or must be checked by staff, and UEA restricts it to plagiarism officers because of "reported inaccuracy".

Is the Turnitin licence dispute the same as the AI-detector decisions?

No. The decisions in this guide are about Turnitin's AI indicator and AI detectors, not Turnitin as a whole: Dundee, Strathclyde and Brunel describe its similarity checking as still in use. The licence question is about a proposed new Turnitin End-User Licence Agreement, which Aberystwyth reported on 5 October 2026 had been postponed until September 2027. If your university has announced a change to its Turnitin contract, read its own announcement.

Sources and further reading

Each university's own page, opened on 17 October 2026, with several rechecked on 18 October (dates as shown on the page or recorded by the tracker):

• King's College London (Oct 2023, updated Apr 2025)
• University of Greenwich (31 Mar 2023)
• University of Birmingham (Jul 2023, updated Jan 2025)
• Durham University (no date on page)
• University of Liverpool (18 Jul 2024)
• University of Dundee: teaching guidance (updated 16 Sep 2026) · student Turnitin guide
• Newcastle University: library FAQ (updated 26 Mar 2025) · Scholarship Insights Newcastle article by a Business School academic (5 Aug 2025)
• Bath Spa University (no date on page)
• University of Bath (published 16 Apr 2026, updated 11 Aug 2026)
• Bournemouth University (13 Apr 2023)
• University of South Wales (PDF, early 2024)
• University of Manchester (no date on page)
• University of Portsmouth (no date on page)
• University of Oxford (16 Feb 2026)
• University of Sheffield (no date on page)
• UCL (first published 8 Oct 2024)
• Queen's University Belfast (PDF, 2025–26 version)
• University of Cambridge: FAQ (no date on page) · assessment guidance
• University of St Andrews (PDF, Jan 2025)
• University of Strathclyde (PDF, version 2.0, Aug 2025)
• University of Reading (PDF, Nov 2023)
• University of Edinburgh (28 Mar 2025)
• Manchester Metropolitan University (no date on page)
• Goldsmiths, University of London (last updated 23 Sep 2026)
• Aberystwyth University (27 Sep 2023) · Turnitin statement (5 Oct 2026)
• Brunel University of London (undated)
• Ulster University (undated)
• Imperial College London (entry dated Sep 2026)
• Edinburgh Napier University (PDF, Aug 2025)
• University of the Arts London (undated)
• Middlesex University (PDF, Sep 2026)
• University of Sussex (undated)
• University of Bristol (20 Nov 2025)
• Northumbria University (PDF, Mar 2024)
• Leeds Beckett University (PDF, 13 Feb 2024)
• Birkbeck, University of London (PDF, 2025–26)
• Abertay University (undated)
• University of East Anglia (PDF, 6 Nov 2023)

Counted from the tracker's August 2026 record, because the page did not open for us in October 2026 or no longer carries the recorded statement:

• University of Nottingham: 4 Apr 2023 · 15 Apr 2024 (no longer publicly accessible)
• University of Warwick (26 Apr 2023) (now asks for a login)
• City St George's, University of London (now asks for a login)
• University of Glasgow (Apr 2023; page no longer available)
• University of Leeds (retrieved by the tracker 23 Aug 2026; page no longer available)
• Birmingham City University (the page has since changed and no longer carries the quoted sentence)

Genutext: UK university AI-detection policy tracker · Does Turnitin detect AI? · How accurate is Turnitin's AI detection? · AI detector scores as evidence in UK universities · AI detectors and non-native English writers · Accuracy methodology

Turnitin is a trademark of Turnitin, LLC. Genutext is independent and not affiliated with, endorsed by, or connected to Turnitin or any university named here. University positions are quoted as published on each university's own pages, read on 17 and 18 October 2026, and they can change. Check the linked page for the current position.