Direct answer
To check for ChatGPT usage, run the text through an AI-detection tool that analyzes writing patterns and returns an AI-writing percentage with flagged segments; Turnitin's AI writing detector does exactly this, reporting an overall score plus sentence-level highlights, and it displays *% instead of an exact figure when AI detection falls below its 20% confidence threshold [1]. A high score is a signal to review, not automatic proof of misconduct, so the result should always be read alongside the draft's revision history and context [1].
How Do You Check a Document for ChatGPT Usage with Turnitin's AI Detector?
Turnitin's AI writing detector works by analyzing the linguistic properties of the submitted text itself, not by searching a database of known ChatGPT outputs. It looks at patterns such as sentence predictability, perplexity, and burstiness, then estimates how likely each segment is to have been machine-generated [2]. That is why the tool can flag text even when no one has ever submitted that exact passage before.
When a document is processed, the detector produces an overall AI-writing percentage and marks the specific segments that triggered the model's attention. Only the text that was actually submitted is analyzed, so the report reflects the draft in front of you rather than a comparison against some external corpus of AI writing [2]. Instructors typically see this report inside their learning management system, and students do not always have default access to it.
Because the analysis is statistical, the output is a probability rather than a verdict. Turnitin itself frames the score as one piece of evidence that should be considered with other signals, and it advises against treating a single number as conclusive [2]. Reading the flagged segments and comparing them with your own drafting process is the practical next step.
What Does the Turnitin AI Writing Report Actually Show, and What Do the Highlighted Flags Mean?
The report presents two layers of information. First, there is an overall percentage summarizing how much of the document the detector considers likely AI-generated. Second, the document view highlights individual segments, each carrying its own indicator, so you can see precisely which sentences or passages drove the score rather than guessing from a single headline number [3]. This segment-level detail is what makes the report useful for revision.
The 20% threshold governs how that number is displayed. When the detector's confidence is below that threshold, Turnitin shows *% rather than an exact percentage, signaling a low-confidence result rather than a confirmed finding [3]. Seeing *% is therefore not the same as seeing a specific high score, and it should not be read as an accusation.
Flags are also not infallible. Human-written prose can occasionally be highlighted, and the use of paraphrasing tools, translation software, or heavy grammar editors can shift the patterns the detector measures [3]. For that reason, a flagged segment is best treated as a prompt to examine your own text and explain your process, not as a final judgment about authorship.
Can You Check Your Draft for ChatGPT Usage Before You Submit It — and Avoid a False Accusation?
Yes, and doing so before the deadline is far safer than discovering a problem afterward. Reviewing flagged text while you still control the document lets you revise passages in your own voice, add specific examples and citations, and replace generic phrasing with evidence drawn from your own reading [4]. That kind of revision improves the writing and reduces the chance that a legitimate draft reads as machine-generated.
It also helps to understand how institutions interpret these reports. Turnitin and most academic-integrity teams treat AI detection as one signal among several, weighed against drafts, notes, and conversations with the student, rather than as an automatic finding of misconduct [4]. Knowing this changes how you should react to a score: the goal is a transparent, well-supported piece of writing, not a number that looks good in isolation.
A pre-submission check gives you the same report style your instructor will see, which removes the guesswork. When you can preview the overall percentage and the individual flagged segments in advance, you can target your edits precisely and walk into submission knowing exactly what the detector responded to [4]. That is the difference between hoping for the best and actually verifying your draft.
If you want that same instructor-facing view of your own draft — the overall AI percentage, the flagged segments, and the similarity summary, all in one place — turnitin0 lets you preview it before the real submission does the talking.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
Get Real Turnitin AI & [Similarity Report](https://www.turnitin0.com/guides/us/turn-it-in-check)
FAQ
Does Turnitin detect ChatGPT specifically?
Turnitin's AI writing detector estimates whether text was machine-generated rather than identifying a particular model, so it does not name ChatGPT in its output. It analyzes writing patterns in the submitted text itself, which means it can respond to ChatGPT-style prose without comparing against a stored library of ChatGPT responses [1].
What does an asterisk percentage mean on the AI report?
An asterisk (*%) appears when the detector's confidence sits below its 20% threshold, indicating a low-confidence result rather than an exact score [3]. It is not a hidden high score; it is Turnitin signaling that the signal was too weak to report precisely.
Can human-written text be flagged as AI?
Yes. Because detection is statistical, human prose can occasionally be highlighted, and tools such as paraphrasing software, translation engines, or aggressive grammar editors can influence the patterns being measured [3]. That is why reports are meant to be reviewed with context rather than treated as automatic proof.
Should I check my draft before submitting it?
Checking beforehand is the safer approach, because you can revise flagged passages in your own voice and add specific evidence while you still control the document [4]. It also lets you see the same report style your instructor will see, so there are no surprises at submission.