The University has received questions from students and staff about whether work submitted to Turnitin is used to train artificial intelligence. This FAQ explains the University of Portsmouth position and distinguishes AI-model training from Turnitin’s established similarity-checking repository. The University does not permit student submissions to be used to train Turnitin AI models.
Turnitin, AI and student submissions: frequently asked questions
Students and colleagues have raised understandable questions about how Turnitin handles submitted work, particularly following discussion across the higher education sector about artificial intelligence. The key point for the University of Portsmouth is that Turnitin’s AI-writing detection functionality is not enabled. Turnitin states that student papers are not used to train generative AI models or large language models. It also states that current submissions are not used to train its AI-writing detection classifier where that functionality is not enabled.
This is separate from Turnitin’s long-standing use of submitted work for similarity checking. Depending on how an assessment is configured, a submission may be retained in a Turnitin repository so that it can be compared with future work. Repository retention and AI-model training are different activities.
No. The University does not enable Turnitin’s AI-writing detection functionality. Turnitin also states that it has not used student papers to train generative AI models or large language models and has no plans to do so. On the University’s current configuration, student submissions are not used to train Turnitin’s AI-writing detection classifier.
Generative AI and large language models create new content in response to prompts. Turnitin’s AI-writing detection feature uses a classifier designed to attempt to identify text that may have been produced or modified using AI. The University does not have that detection feature switched on and never has. Turnitin’s similarity report, which the University does use, is a different service: it compares submitted work with other sources and highlights matching text for academic review.
The University uses Turnitin to support electronic submission, similarity checking, marking, grading, feedback and academic-integrity review. A similarity report is not, by itself, a finding of misconduct. It provides information that must be interpreted in context by an appropriate member of staff.
A full submission may be retained in Turnitin’s repository so that it can be compared with future submissions. The correspondence reviewed for this article indicates that repository retention can continue unless the University requests deletion or the assignment is configured not to add the work to a repository. This storage supports similarity checking and should not be described as AI-model training.
Students with a concern should raise it with their module team as early as possible. Depending on the assessment and available configuration, a module team may be able to consider an institutional-repository setting or a Moodle submission route without similarity checking. These options can affect the scope of the similarity report and the marking workflow, so they are not automatic and must be considered in the context of the assessment requirements and University processes.
Requests about a specific submission should be raised with the relevant module team or University support route. The University can then consider the request against the assessment record, academic-integrity requirements, retention obligations and the available Turnitin deletion process. Do not contact Turnitin directly about a University-managed submission unless advised to do so.
Within the University’s current use, submissions support the assessment workflow, including electronic submission, similarity reporting, marking, feedback and related academic-integrity review. Submitted work may also be retained for future similarity comparisons. The University has not enabled Turnitin’s AI-writing detection feature, and current submissions are therefore not used to train that classifier under the position described above.
The University’s use is governed by its institutional contractual arrangements with Turnitin, including the Master Registration Agreement, the relevant order documentation and service terms, together with the End-User Licence Agreement and associated privacy and data-processing provisions. The precise contractual document set and precedence should be confirmed by the University’s procurement and information-governance specialists if a formal or legal response is required.
For an assessment-specific question, contact the module team first. Staff seeking advice about assessment setup or digital learning practice can contact CADI through the usual University support route. Formal requests about personal data, contractual terms or information rights should be directed through the relevant University information-governance or information-rights process.
In summary
- The University does not have Turnitin’s AI-writing detection feature switched on.
- Turnitin states that student papers are not used to train generative AI models or large language models.
- On the current University configuration, new submissions are not used to train Turnitin’s AI-writing detection classifier.
- Submissions may still be retained and compared for similarity checking. This is separate from AI training.
- Students and staff should raise assessment-specific concerns through the module team, with specialist University advice used for formal data-protection or contractual questions.
- Further information can be found on the Turnitin Trust Centre website: https://www.turnitin.com/trust-center