Direct answer
For teachers, a good AI detector is one that is accurate, keeps false positives low, and produces evidence you can actually act on. Turnitin's AI writing detection adds an AI writing indicator to the Similarity Report that shows the overall percentage of a document that may have been generated or modified by AI tools, with sentence-level highlights in a linked report [1]. Only instructors and administrators can see this indicator, which makes it a practical starting point for reviewing student work. This guide explains how accurate these detectors really are, what the report shows, and how to review a real Turnitin AI report before you respond to a student.
How Accurate Are AI Detectors at Catching AI-Written Essays?
Accuracy is the first thing teachers ask about, and it depends on both the tool and the document. Turnitin's detector segments a submission's sentences into overlapping sections, scores each section from 0 to 1, and pools those scores into a single document-level percentage [2]. Its model is trained to recognize content from GPT-4o, the GPT-5 series, Gemini, Claude, LLaMA, Mistral, DeepSeek, and tools built on those models [1].
False positives are the real concern, because wrongly accusing a student is far more damaging than missing a flag. Turnitin states that it maintains a false positive rate below 1% for long-form English documents [1]. At the same time, it acknowledges that detection is less reliable for short or unconventional writing, which is why percentages between 1% and 19% are deliberately displayed as an asterisk (*%) so teachers do not over-interpret them [2].
Turnitin is explicit that the percentage should not be used as the sole basis for grading or disciplinary action, and that the model may misidentify human-written, AI-generated, and AI-paraphrased text [2]. In practice, a good detector for teachers is one that treats its output as evidence to investigate rather than a verdict [2]. Pair the AI percentage with the similarity score and a conversation with the student, and you have a defensible workflow.
What Does a Turnitin AI Writing Report Show Teachers?
The report teachers open in the grading interface is not a vague flag but a structured document. It shows an overall percentage of the document's qualifying prose that the model predicts could be AI-generated or AI-modified, a Submission Breakdown bar, and blue highlights on the specific sentences that were flagged [2]. Understanding where the report comes from matters: students generally cannot self-check their work first, because unless the institution has enabled Turnitin Draft Coach or the instructor has set up a practice assignment that allows resubmissions, a student's first upload is often the final one [3].
The AI writing indicator is visible only to instructors and administrators, not to students [1]. Scores between 1% and 19% appear as an asterisk (*%) because false positives are more likely in that band, while 0% and scores of 20% or higher are shown as numbers [2]. The AI percentage is independent of the similarity score, and AI highlights do not appear inside the Similarity Report itself [2].
Because students usually cannot preview the report, institutional settings shape the entire checking experience: Draft Coach availability and resubmission allowances determine whether a student can ever see a report before the final submission [3]. As a teacher, reviewing the AI percentage alongside the similarity score and the highlighted sentences gives you the complete picture you need before talking to the student [3].
How Can Teachers Review a Real Turnitin AI Report Before Responding to a Student?
Before emailing a student about an AI flag, review the actual report rather than acting on a single number. Turnitin's guidance for educators is that AI writing detection supports, but never replaces, professional judgment, and that the score should not be the sole basis for an academic-integrity decision [4]. A responsible review starts with the per-sentence highlights so you can see exactly which passages the model flagged and judge whether they look like AI output [4].
Compare those highlighted passages against the similarity report and the student's own writing style, then prepare specific questions for a conversation rather than a verdict [4]. When the report from your LMS is not yet available, you can still review a real Turnitin report by running the document through a checking service such as turnitin0, which returns the same AI writing indicator, flags, and similarity summary that instructors see in institutional systems. Having the concrete report in front of you makes the conversation with the student fair and specific, and it protects both of you from decisions based on hunches [4]. A good detector for teachers is therefore not the one with the scariest alerts, but the one whose evidence you can actually act on and defend [4].
Reviewing a real Turnitin AI report should not require waiting for a flagged grade to surface. With turnitin0, you can upload a student draft as a.docx,.pdf, or.txt file and receive the actual Turnitin AI writing report and similarity summary in minutes — the same report format instructors see in their academic systems, complete with the overall AI score and per-sentence flags. See what a real report looks like before you have that conversation.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is Turnitin's AI detector accurate enough for teachers to trust?
Turnitin reports a false positive rate below 1% for long-form English documents and detects output from major LLMs including GPT, Gemini, Claude, and DeepSeek [1]. Accuracy is lower for very short or unconventional texts, which is why scores between 1% and 19% display as an asterisk [2]. Treat the score as strong, but not conclusive, evidence.
What does the asterisk (*%) mean in an AI writing report?
When the AI writing percentage falls between 1% and 19%, Turnitin shows an asterisk instead of a number because false positives are more common in that band [2]. Only 0% and scores of 20% or higher appear as numeric values [2]. Teachers should review the highlighted sentences before drawing any conclusion.
Can students see the AI writing report before submitting?
Generally no. Students can only check their work beforehand if the institution has enabled Turnitin Draft Coach or the instructor has created a practice assignment that allows resubmissions [3]. Otherwise the first upload is typically the final one, so the report a teacher reviews is generated from the official submission [3].
How should teachers use an AI detection score when responding to students?
Use the report as a conversation starter, not a verdict. Turnitin is explicit that the percentage should not be the sole basis for action, and educators should apply their institution's policy and professional judgment alongside the report [1][2]. Reviewing the flagged passages with the student is the recommended next step [4].
What file requirements apply to AI writing detection?
Documents must be under 100 MB, contain between 300 and 30,000 words of prose in a long-form format, and be written in English, Spanish, or Japanese [2]. Non-prose content such as poetry, scripts, code, bullet points, and tables is not reliably detected [2].