Turnitin Trust Center
Protecting your data
You’ve trusted us with your data for nearly thirty years. We recognize the great responsibility that comes with that trust and that the introduction of AI to education brought more questions and concerns about data use and privacy.
We are building on the decades of trust you’ve given us with a transparent approach to product design and data use. Learn about how we are building solutions with AI, how we use and protect your data, and our privacy policy here in our trust center.
Turnitin’s AI principles
We build solutions that support learning integrity and help educators navigate AI integration through transparency and responsible use.
How we design and deliver AI features is guided by five core principles:
- Fairness and Inclusion. Design and test AI so it works equitably across diverse users and contexts. Avoid bias, monitor for disparate impact, and keep humans in the loop.
- Transparency and Human Centered Design. Make AI understandable. Include clear disclosures and plain language explanations of what it does and its limits.
- Privacy, Safety, and Security. Treat data used in AI systems with the same care as all other personal data. Limit use, secure access, and apply techniques like anonymization or pseudonymization
- Learning and Adaptation. Review, test, and refine AI systems regularly. Stay current on emerging research, feedback, and legal standards.
- Accountability and Governance. Maintain documentation, risk assessments, and cross-functional reviews so ownership and oversight are clear at every stage.
AI definitions at Turnitin
Defining terms as they relate to Turnitin products
Classifier vs. Generative AI models:
- Classifier AI models categorize or label existing data, returning a numeric prediction about AI writing – they do not write, rewrite, or produce text.
- Generative AI models create new content including text, images, audio, code, video, etc., based on learned patterns and typically, from a prompt.
We do not train any generative AI models, or LLMs. We only train a classifier model for AI detection in our Turnitin Originality license.
Training and Evaluating AI models: AI models are trained on how to work, and then evaluated on how well they work. These two definitions are essential to understanding how we do and do not use student data.
- Training: Training our classifier AI model for AI detection in our Turnitin Originality license means teaching the model what is human writing and what is AI writing. If you do not have Turnitin Originality with AI detection enabled, your current student data is not used for training our classifier AI model. If you do have Turnitin Originality with AI detection enabled, your current student data is anonymized and may be used to train our classifier AI model.
- So how does training work?
- We train our AI model on both AI-generated text and authentic, anonymized academic writing. The training dataset includes a representative sample of data across geographies and subject areas. We take into account statistically under-represented groups such as second-language learners, English users from non-English speaking countries, students at colleges and universities with diverse enrollments, and less common subject areas including anthropology, geology, sociology and others to minimize bias.
- The human writing is curated from authorized databases, anonymized and pseudonymized.
- After training the classifier model, the datasets stay in secure storage. The model is deployed on servers Turnitin controls. The interface is made available to customers, not the data.
- So how does training work?
- Evaluating: Evaluating our classifier AI model in our Turnitin Originality license means assessing its effectiveness and false positive rate. We do evaluate our AI model against current student writing so we know if it works, and it never leaves our Turnitin systems.
AI in our products
Transparency is a critical part of our AI principles. Below is a list of AI or AI-powered features so institutions, educators, students, publishers, and researchers can see how AI is used within our products.
This list is of publicly available products and features and may include beta or preview programs for licensed customers, where noted.
Last updated: 23 September 2026
| Turnitin Solution | Features | Description | AI service provider / service | Notes/further information |
| Turnitin Feedback Studio | Similarity Report - Match Group categorization | Natural language processing (NLP) technology is used to recognize in-text citations and quotation marks associated with matched text, helping instructors easily recognize the matches where the student attempted to include proper attribution versus those where they may have copied text without attribution. | Turnitin - in-house | |
| Video feedback closed captioning | AI is used to transcribe voice to text. | Amazon Transcribe | ||
| Translated matching | Machine learning technology is used to establish semantic similarity between texts, helping detect content matches across a broad range of languages. | Turnitin - in-house | ||
| First pass feedback & summarization (in beta) | Uses AI to provide contextually aware feedback suggestions to instructors based on student submissions, instructions and rubric. It also takes all the feedback applied to the student’s paper, whether that originated from AI or the instructor themselves, and the rubric scoring, to draft and format an initial summary comment for teachers to review. | Claude Haiku 4.5 | Anthropic’s pre-trained models do not use or retain any student or assignment data submitted to Turnitin for AI training and evaluation | |
| iThenticate | AI writing detection (available as an add-on) | Using an LLM-powered text classification model, this solution helps educators identify text that has likely been generated and/or modified by AI writing tools. | Turnitin - in-house | AI writing detection FAQs |
| Similarity report - Match Group categorization | Natural language processing (NLP) technology is used to recognize in-text citations and quotation marks associated with matched text, helping reviewers easily recognize the matches where the author attempted to include proper attribution versus those where they may have copied text without attribution. | Turnitin - in-house | ||
| Turnitin Similarity | Similarity Report: Match Group categorization | Natural language processing (NLP) technology is used to recognize in-text citations and quotation marks associated with matched text, helping instructors easily recognize the matches where the student attempted to include proper attribution versus those where they may have copied text without attribution. | Turnitin - in-house | |
| Turnitin Originality (available as an add-on to Turnitin Similarity, Turnitin Feedback Studio & Originality Check) | AI writing detection | Using an LLM-powered text classification model, this solution shows text that has likely been generated and/or modified by AI writing tools. | Turnitin - in-house | AI writing detection FAQs |
| Authorship | Authorship uses computational linguistics and natural language processing to establish if the submission's writing is consistent with the student's prior work. | Turnitin - in-house | ||
| Turnitin Clarity (available as an add-on to Turnitin Feedback Studio, requires Turnitin Originality) | AI assistant | Backed by an AI-powered LLM, the AI assistant provides feedback and guidance to students (without actually writing for the student or changing their content), helping them use AI responsibly and improving their writing skills. | Anthropic's Claude 3.5 Haiku in North America and Claude 3.5 Sonnet in Europe and the Asia Pacific regions. | Anthropic’s pre-trained models do not use or retain any student or assignment data submitted to Turnitin for AI training and evaluation. We do use anonymized student assignment data to evaluate, improve, and extend the quality of the Turnitin AI assistant responses and capabilities. Data anonymization tools remove personal identifying information from customers’ data prior to it being used for feature improvements. |
| Citations checker | AI is used to check for formatting of in-text citations and works cited. It does not automatically change text. | Claude Haiku 4.5 | ||
| Grammar checker | AI is used to provide grammar feedback to students without modifying the original meaning of the text. | Amazon Nova Lite |
What you can control
AI writing detection
If your institution is licensed for Turnitin Originality and does not want submissions to be processed for AI writing detection, it can be disabled from the administrator account settings page.
Private repositories
Administrators may configure private repositories, which store submissions separate from the standard Turnitin database and are searched against only by the institution’s users. Content added to a private repository are not used for AI writing training.
Clarity AI assistance
The AI assistance tools within Turnitin Clarity can be enabled or disabled by individual instructors. The pre-trained LLMs do not use or retain any student or assignment data submitted to Turnitin for AI training and evaluation.
Privacy policy
View our full privacy policy.