Direct answer
To check whether a piece of writing is AI-generated, run the document through an AI writing detector such as Turnitin, then read the report rather than the headline number: the overall AI score tells you what share of the text was flagged, and the highlighted segments show exactly which sentences triggered it. Turnitin's detector analyses submitted documents and returns a percentage alongside colour-coded passages, and it displays a *% placeholder instead of a precise figure when the signal falls below its confidence threshold [1]. Treat the result as evidence to interpret, not a verdict — formulaic, heavily templated, or non-native-English prose can be flagged even when a human wrote every word [1].
How do I check if a piece of writing was written by AI?
The practical workflow has four moves: isolate the text, submit it to a detector that reports at sentence level, read the flagged segments, and then judge those segments yourself. Start by getting the text into a supported file format — detectors generally operate on uploaded documents rather than on a paragraph pasted into a chat box, because formatting, headings, and reference lists all feed into how the model segments the prose [2]. Keep the document intact; stripping citations or reformatting can change which passages get flagged.
Second, choose a detector whose output you can actually audit. A bare percentage is close to useless on its own, because it tells you a proportion without telling you where the machine-generated signal sits [2]. What you want is a report that pairs the overall figure with per-segment highlighting, so you can look at the specific sentences and ask whether they genuinely read as machine-written or whether they are simply conventional academic phrasing [2].
Third, read the score in context. Turnitin's AI writing indicator is designed as a signal for instructors to investigate, not as an automatic misconduct finding, and Turnitin states plainly that a high score is not proof of misconduct on its own [2]. That framing matters for how you use the result: if you are checking your own draft, a flagged passage is a prompt to revise, not a reason to panic.
What do Turnitin's AI detection score and highlighted flags actually mean?
The headline percentage is a share of qualifying text, not a confidence rating. Turnitin reports the proportion of the submitted document that its model identifies as AI-generated, and the report then highlights the specific segments that contributed to that figure, so the number and the highlighting should always be read together [3]. A document can show a moderate overall score driven entirely by two flagged paragraphs, with the rest of the text clean.
The *% display is the detail most students misread. When the detected signal falls below Turnitin's confidence threshold, the report shows an asterisk in place of an exact percentage rather than a low number, because the model is not confident enough to commit to a figure [3]. Seeing *% is therefore a low-confidence outcome, not a hidden high score and not a rounding error — it means the detector found something, but not enough to quantify reliably [3].
Highlighted flags carry the same caveat. Turnitin notes that its detection can produce false positives, particularly with highly formulaic writing, text produced under heavy template constraints, or writing from authors working in a second language, and it advises that flagged segments be reviewed by a human before any conclusion is drawn [3]. Read the highlights as a list of passages worth a second look, then apply your own judgement to each one.
Can I check my own draft with Turnitin before I submit it?
Institutional Turnitin access normally runs through a course assignment, which means a self-check can end up inside the same submission workflow your instructor sees — so students often look for a non-repository check that returns the same style of report without lodging the file in the institutional paper database [4]. That distinction is the one worth understanding before you upload anything: a repository submission adds your document to a comparison corpus, while a non-repository check does not [4].
What you should expect from any legitimate pre-submission check is a report that mirrors the institutional output — an overall AI score, segment highlighting, and a separate similarity figure — because a check that returns a different format teaches you nothing about what your instructor will see [4]. Turnitin's own guidance for students emphasises understanding how the report is produced and what the indicators mean, rather than chasing a number, which is the right posture for a self-check too [4].
For a draft that matters, run the check early enough to act on it. If the report comes back clean, you have confirmation and can submit with confidence; if segments are flagged, you have time to rewrite those passages in your own voice and re-check before the deadline rather than after it [4].
Reading the report is the easy half. The harder half is knowing whether the report you are reading matches the one your instructor will open — which is exactly the gap turnitin0 closes, by returning the same Turnitin AI and similarity reports, with the same score display and the same segment highlighting, on your draft before it is ever submitted.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is a high Turnitin AI score proof that a document was written by AI?
No. Turnitin presents the AI writing indicator as a signal for human review, and its guidance notes that false positives occur with formulaic or template-constrained writing [2]. The score tells you where to look, not what to conclude.
What does an asterisk (*%) instead of a percentage mean?
It means the detected signal fell below Turnitin's confidence threshold, so the report withholds an exact figure rather than reporting a low one [3]. It is a low-confidence result, not a concealed high score.
Can checking my draft get it added to Turnitin's paper database?
Only if the check is a repository submission. A non-repository check does not add your file to the student paper database, which is why students choose that route for pre-submission checks [4].
Does rewriting flagged passages actually change the score?
Often, yes — the detector responds to the specific segments it highlighted, so revising those passages in your own phrasing and re-checking gives you a direct before-and-after comparison [3]. Re-checking is the only way to confirm the revision worked.
How long should I leave between checking and submitting?
Enough time to act on the report. Running the check days rather than minutes before the deadline means a flagged result is a revision task instead of a crisis [4].