Plagiarism checking software Turnitin has come under fire for its new proposed legislation that would potentially see students’ work used to train new AI services, prompting multiple UK universities to terminate their contracts.
Universities such as Southampton, Newcastle and Westminster have publicly stated their intentions to terminate relationships with Turnitin. Similarly, Cambridge has refused to sign Turnitin’s new End User Licensing Agreement (EULA), which would see the new AI training implemented.
The proposed changes to the EULA were initially set for July, but have been postponed to September 2027 due to the understandable outrage from students and staff. In a statement, Turnitin said they “do not train a generative AI model, or LLM,” but “do train a classifier model, which identifies likely AI-written text and does not produce any content, to predict what is AI versus authentically human.”
Sheffield has currently not publicly commented on the Turnitin controversy, much less publicly made plans to depart from the software. According to the academic integrity section of the University’s website, they do not use generative AI detection tools “due to concerns over their error rates and the potential for both false positives and false negatives.” The University does, however, use Turnitin to detect plagiarism.
When asked about the University’s stance on the controversy, a spokesperson for Sheffield told Forge Press that they were “aware of the recent news about Turnitin,” and that “The University is continuing to work with higher education sector groups to monitor future updates to Turnitin’s privacy policies.”
They also told us: “Turnitin has confirmed that student and staff data is never used to train AI models.” This would change if Sheffield continues to use the plagiarism detection software when the new EULA is implemented.
Concerns about AI usage in education, in multiple ways, are not new. In response to questioning about Turnitin’s new proposal, 98% of students understandably said they wouldn’t feel comfortable having their work used to train AI. One student said: “AI was never a choice, it just appeared, replaced everything and demanded us to be okay with it.”
Submitting AI-produced work is prohibited by the University. Students are encouraged to use it responsibly, and as a “powerful tool to support your learning.” On Sheffield’s website, they say that they’re “aiming to support students and staff to take a positive approach to the ethical and responsible use of GenAI.”
If a student is found to have potentially used AI, they “may be called on to verbally explain ideas present in [their] work.” This creates a grey area, putting students in situations where they may be falsely accused.
When asked by Forge Press, 11% of students said they’d been falsely accused of using AI. Though this is a small amount, false accusations can cause heavy stress for students, due to the long process of getting cleared.
One student studying Music was falsely accused of academic misconduct, with the University claiming that written work appeared to have been created using “text generators / generative intelligence.” They spoke about the experience to Forge Press.
They said: “I was initially reached out to by email from the department admin with my lecturer, tutor and head of the office at the time copied in. [..] I was expected to answer questions about my write-up in a meeting to confirm my knowledge of what I was writing.”
They were then asked to attend an online meeting. “At this meeting, I was asked a series of questions about how I structured my piece as well as the contents of it, all of the information of which I knew because I wrote it and did not use AI.
“I explained everything accurately that was taken issue with and felt as though just because I lacked complete accuracy in my referencing and the text style changed in the marking tool that these things were being used as a basis to make the accusation.”
After the meeting, they received a follow-up email stating that they would not penalise them, but that “a record of the incident [would] be retained on [their] university file” and that the confirmation email “should be treated as a formal warning.”
The student told us that they felt “a mix of emotions” during the experience. “It may just be my nerves but the feeling of being suspected and accused weighed heavily on me, though I would say that while intimidating they in the email did not jump to any conclusions about which way it went – it was simply the accusation itself that felt so out of place.
“The main feeling I remember having was a disappointment of the incompetence of the system. [..] I felt really puzzled as to how far the reach was to mean that these things led to my usage of AI, and the closing remarks of the meeting generally followed the gist that my write-up was of ‘poor enough quality that it looked like AI’ and that itself was said not to be a good thing. This made me feel like on top of being accused I was being insulted for my work which felt belittling and petty, or perhaps defensive.”
The student concluded by saying that their “faith in the system is lower,” and that methods universities may use to crack down on AI would give them less assurance as opposed to more assurance.”
As aforementioned, the global crackdown on students using AI creates a debate with no clear answer:
How do universities ensure AI usage is fair to maintain academic integrity, whilst also avoiding false accusations that can cause unnecessary stress?
A report by Dr David Grundy at Newcastle Business School concluded that Turnitin’s AI detection software ranged in inaccuracies, with data from Temple University finding that it was only 77% accurate in detecting AI, and had a 7% misflag rate for human-written writing. Though Sheffield clarified that they don’t use Turnitin as an AI detection tool, it still poses questions about just how accurate the software is about detecting different forms of plagiarism.
The same study also makes another good point: if work is flagged as having been produced using AI, investigators may fall victim to confirmation bias, meaning they may overlook work as “too good to be the student’s own” whilst not taking into account a student’s personal abilities.
This puts international students at risk in particular, who may use Google or grammar tools to aid them in their work. In July 2025, the Office of the Independent Adjudicator published four case studies involving students falsely accused of AI usage – three of those were international or second-language students.
Similarly, a study published by Stanford University found that over 61% of essays written by non-native english speakers were falsely flagged as AI generated, in comparison to near-accurate results for native speakers.
The newly-proposed Turnitin legislation proves that AI isn’t leaving the education sector anytime soon, raising an urgent question regarding how the University navigates the ever-growing need for guidance on navigating academic integrity in an AI age: in the pursuit to protect academic integrity, are students’ integrity being protected too?
