Direct answer
Yes, Turnitin's AI writing detector does identify ChatGPT-generated text, but it reports a probability-style indicator rather than proof of authorship. Turnitin analyses submitted work for patterns associated with AI generation and returns an overall percentage of qualifying text flagged as AI-written, which is why ChatGPT drafts are routinely surfaced in reports [1]. The same documentation is explicit that the result is a signal for human interpretation, not a verdict, and that short, heavily edited, or mixed passages are the weakest cases [1]. So the honest answer is "yes, it detects — and no, it does not decide."
How Does Turnitin Detect ChatGPT Text, and How Accurate Is It?
Turnitin's detector does not look for a watermark or a hidden signature inside ChatGPT output. It evaluates the statistical and structural regularity of the prose itself, working at sentence level across the qualifying text in the document, then reports the share of that text it considers likely AI-generated [2]. Because the analysis is sentence-level rather than document-level, a report can flag three paragraphs of a twelve-page essay and leave the rest untouched — a pattern that confuses students who expect an all-or-nothing score [2].
Accuracy is where the "really" in the question earns its scepticism. Turnitin positions the indicator as one input in an academic judgement process, not a standalone finding, and its own guidance stresses that human review is required before any conclusion about misconduct [2]. Detection confidence drops when text is short, when it has been paraphrased or restructured by hand, and when human and AI writing are interleaved at sentence level [1]. In practice this cuts both ways: a clean ChatGPT draft is usually caught, while a heavily rewritten one may score low even though it began as generated text [1].
The practical takeaway for students is that "will it be detected" is the wrong framing. The detector is consistent enough that submitting unedited ChatGPT prose is a genuine risk, and imprecise enough that a flagged score is not automatically an accusation [2]. What matters is knowing what your own draft produces before someone else runs it.
What Does a Turnitin AI Score Actually Mean, and Why Does It Sometimes Show an Asterisk?
The headline figure in a Turnitin AI report is the percentage of qualifying text the detector attributes to AI generation — not a grade, not a certainty, and not a measure of how much of the essay "feels" machine-written [3]. It is calculated only over text long enough to analyse, so reference lists, short quotations, and fragmented headings typically fall outside the calculation and can make the percentage look inconsistent with the visible page count [3].
The asterisk is the detail most students misread. When the detected AI share sits below Turnitin's confidence threshold, the report displays an asterisk in place of the exact number, because the signal is too weak to present as a precise percentage [3]. An asterisk therefore means "low-confidence, low-level signal" — it is not a confirmed zero, and it is not a clean bill of health [3]. Treating it as proof of innocence is the single most common misreading of the report.
Reading the score correctly also means reading the highlighted passages alongside it. A 40% figure concentrated in one section tells a very different story from 40% scattered evenly across every paragraph, and instructors are trained to look at distribution rather than the headline number [3]. If you only ever see one number and never the highlighted text, you are missing the part of the report that actually carries the evidence.
Can a Turnitin AI Detection Report Be Checked Before Final Submission?
Yes — an independent pre-submission check returns the same two reports your institution would generate: an AI detection report and a similarity report, delivered as downloadable PDFs before you submit anything [4]. The purpose is diagnostic rather than defensive: you see the percentage, the asterisk if there is one, and the specific passages flagged, while you still have time to revise the text that triggered them [4].
The mechanism matters as much as the result. A non-repository check means the file is not added to Turnitin's student paper database and the reports are not shared with third-party databases, so a pre-check does not create the very match your instructor would later see [4]. That distinction is why pre-checking is a legitimate revision step rather than a workaround, and it is the reason the service is built around documents being deletable from your account afterwards [4].
Timing is the practical constraint. Turnaround is under 15 minutes in the large majority of orders, with most completing in 5 to 15 minutes and a guaranteed ceiling of 30 minutes during queue spikes, which is short enough to fit inside a normal revision cycle before a deadline [4]. For a student holding a ChatGPT-assisted draft, that window is the difference between guessing at your risk and reading it.
If the question you actually need answered is "what will my own draft score," that answer only exists as a report — and turnitin0 produces the real Turnitin AI detection and similarity PDFs, with the flagged passages highlighted, before your instructor ever sees the file. Independent of Turnitin, LLC, and used by more than 20,000 students across the US, UK, Canada, Australia, New Zealand, and Ireland, it turns a guess about detection into a document you can read, revise against, and resubmit with confidence.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does Turnitin detect ChatGPT with 100% accuracy?
No detector is 100% accurate, and Turnitin does not claim to be — it presents the AI indicator as a signal requiring human review [2]. Unedited ChatGPT prose is usually flagged, while heavily paraphrased or interleaved text is where confidence drops [1].
Why does my Turnitin AI report show an asterisk instead of a number?
An asterisk appears when the detected AI share falls below Turnitin's confidence threshold, meaning the signal is too weak to display as a precise percentage [3]. It is a low-confidence indicator, not a confirmed zero.
Can Turnitin tell if I edited ChatGPT text by hand?
Partially. Sentence-level analysis means hand-edited passages that retain machine-like structure can still be flagged, while thorough restructuring tends to reduce the detected share [2]. There is no reliable rule for how much editing is "enough" — the report is the only way to know.
Does a pre-submission check add my paper to the Turnitin database?
No. A non-repository check does not add the file to Turnitin's student paper database and does not share reports with third-party databases, and files can be deleted from the account afterwards [4].
Will my instructor see that I ran a pre-check?
No. The pre-check is independent of your institution's submission, so your instructor only sees what you submit through the course itself [4]. The value is in revising the flagged passages before that submission happens.