Direct answer
Yes, Turnitin can still detect AI-generated text, and its detection model is continuously updated to keep pace with newly released large language models. When a paper is submitted, Turnitin scores the prose sentence by sentence and reports an overall percentage of text it predicts was written or modified by an AI tool, with the flagged passages highlighted for the instructor [1]. Detection works today — but it works as a statistical indicator designed to guide human review, not as an automated proof of misconduct [1].
How Does Turnitin's AI Writing Detection Work?
When a submission is processed, Turnitin extracts the qualifying prose sentences and divides them into overlapping segments for analysis. Each segment is run through the AI detection model and assigned a probability between 0 and 1 that the text is AI-generated rather than human-written [2]. Because the segments overlap, individual sentences can receive multiple scores, which are pooled into one sentence score, and those sentence scores are then aggregated into the document-level percentage shown at the top of the report [2].
The percentage refers only to "qualifying text" — prose sentences in long-form writing such as essays, dissertations, and articles. Poetry, scripts, code, bullet points, and other short-form writing fall outside the model's reliable scope, and Turnitin notes that the AI percentage is fully independent of the similarity score, so a paper with zero plagiarism can still carry a high AI-writing percentage [2].
Results are presented through indicator states: 0% when nothing is flagged, an asterisk (*%) bucket for scores between 1% and 19% (suppressed because sub-20% results are statistically less reliable), and a numerical percentage for 20–100% [2]. The interactive submission breakdown bar then highlights the exact passages that triggered the flag, including text that may have been further modified by AI paraphrasing, word-spinning, or bypass tools [2].
Can AI-Generated Text Still Slip Past Turnitin's Detector?
Turnitin positions its detection explicitly against the evasion arms race: the AI writing detection built into Turnitin Originality includes likely AI-generated content that has been further modified by an AI paraphrasing or bypassing tool, so running AI text through a humanizer does not automatically clear it [3]. The indicator surfaces an overall percentage of likely AI text in the submission, and the detection capabilities are designed to address AI writing, AI paraphrasing, and AI bypassing — the three ways students typically try to hide AI use [3].
That said, no detector is infallible. Turnitin itself cautions that its model may not always be accurate and may misidentify human-written, AI-generated, and AI-paraphrased text, which is why the report is not meant to be the sole basis for academic action [2]. The company reports a false positive rate below 1% for documents of 300+ words and, in July 2026, consolidated its multi-model ensemble into a single updated architecture to keep the detector current [1].
Practically, the odds of slipping through depend on the document and the tool used. Detection covers a wide, continuously expanding list of models — including GPT-4o and later GPT-5.x releases, Gemini, Claude, LLaMA, Mistral, DeepSeek, and Grok — so using a brand-new model does not make output invisible [1]. Because the result is a probability, borderline documents exist, and instructors are instructed to treat the indicator as a conversation starter rather than a verdict [3].
How Can I Check My Draft for AI Writing Before I Submit It?
Here is the catch for students: Turnitin does not allow you to self-check a paper inside Turnitin unless your institution has enabled Turnitin Draft Coach, which lets students run Similarity, citation, and grammar checks inside Google Docs or Microsoft Word before submitting [4]. Without Draft Coach, your ability to preview results depends entirely on your instructor's assignment settings, and there is no built-in student portal for running checks on demand [4].
Without Draft Coach, the only official path is submitting to an assignment that allows resubmissions: the first three attempts generate a Similarity Report immediately, but after three resubmissions within 24 hours you must wait until the next calendar day [4]. If your instructor has disabled resubmissions, your first attempt is final, and you will not be able to preview anything beforehand [4].
The deeper problem is visibility: even when resubmissions exist, the AI writing indicator itself is only visible to instructors and administrators — students rarely see their own AI percentage at all [1]. In most institutional setups, a student discovers their AI score only after it has already affected their grade, which is why pre-submit checking is such a common pain point [4].
Understanding how Turnitin's detector works is only half the battle — the harder part is knowing your own draft's AI score before your instructor ever sees it. Turnitin's own documentation confirms that students usually cannot preview the AI indicator themselves, which leaves a genuine gap in the final hours before submission. turnitin0 closes that gap: upload your draft and get the real Turnitin AI writing and similarity reports your professor would see, with the flagged passages clearly highlighted, so you can revise before the grade is on the line.
※ Turnitin0.com - Turnitin AI Detector Trusted by 20,000+ Students Worldwide
FAQ
Can Turnitin detect text from ChatGPT, Claude, or Gemini?
Yes. Turnitin's English detector is trained on output from major large language models — including GPT-4o and later GPT-5.x releases, Gemini, Claude, LLaMA, Mistral, DeepSeek, and Grok — and it is expanded as new models are released [1].
Can I hide AI text from Turnitin by running it through a paraphraser?
Not reliably. Turnitin's detection capabilities explicitly include AI paraphrasing and bypassing detection for English submissions, so text modified by humanizer or word-spinner tools can still be flagged [3].
What does the asterisk (*%) in the AI report mean?
Scores between 1% and 19% are displayed as *% rather than a number because sub-20% results are statistically less reliable and prone to false positives; only 0% and scores of 20% or higher are shown numerically [2].
Can students see their own AI score before submitting?
In most setups, no. The AI writing indicator is visible only to instructors and administrators, and students can self-check only if their institution enables Turnitin Draft Coach or the assignment allows resubmissions [4].
Is a high AI percentage proof of cheating?
No. Turnitin states that its model may not always be accurate and that the report should not be used as the sole basis for action; instructors apply further scrutiny, academic policy, and human judgment [1].