Direct answer
An AI signals review is the process of reading the AI-detection layer of a Turnitin-style writing report: the overall AI percentage plus the individual highlighted passages ("signals") that the detector flagged as possibly machine-generated. The review matters because a signal is an indicator, not a verdict — Turnitin itself presents the AI percentage as a starting point for human judgement rather than proof of misconduct [1]. A useful review therefore asks three things: what the signals are, how the detector decided to flag them, and whether you can inspect those signals on your own draft before you submit.
What Are AI Signals in a Turnitin AI Writing Report?
An AI signal is a specific passage of your text that the detector's model has scored as likely AI-generated, surfaced inside the AI writing layer of the report rather than in the similarity layer [2]. When you open the report, the overall AI percentage sits alongside highlighted segments, and each highlighted segment can be expanded so the reader can see exactly which sentences produced the signal [2]. This is why the phrase "AI signals review" is more precise than "AI score check": the score is a summary, while the signals are the evidence underneath it.
It also helps to separate the two report types students often conflate. The similarity report compares your text against a database of sources, while the AI writing report estimates how much text looks machine-generated — they use different models and answer different questions [2]. A high similarity match does not create an AI signal, and a clean similarity score does not clear an AI flag.
Finally, access matters. In institutional Turnitin, the AI writing report is visible to instructors by default, and students are not automatically shown the AI percentage [2]. That asymmetry is the single biggest reason students search for an AI signals review in the first place: they know a signal exists somewhere in the system but cannot see it.
How Does Turnitin Review AI Signals and Decide What Counts as Flagged?
Turnitin's detector segments submitted text and estimates, segment by segment, the likelihood that the wording was produced by a large language model, then aggregates those segment estimates into the overall percentage shown at the top of the report [3]. The flagged layer you review is therefore a set of confidence estimates, not a set of accusations — which is exactly how Turnitin frames the output for instructors [3].
The most misunderstood part of the review is the asterisk. Turnitin displays *% instead of a number when the AI score falls below its 20% confidence threshold, meaning the detector found only low-confidence signals rather than a confirmed result [3]. Students frequently read *% as "zero" or as "caught," when it actually means the evidence was too weak for the detector to commit to a figure [3].
False positives are a documented limitation, and they cluster around predictable writing: heavily templated academic phrasing, formulaic transitions, and text that has been rewritten so many times it loses the irregularities of natural drafting [3]. A careful AI signals review therefore treats a flag as a prompt to look at the passage in context — does the flagged sentence reflect your own reasoning, your own citations, and your own voice? — rather than as a final judgement about authorship [3].
Can Students Check Their Own AI Signals Before Submitting?
In most institutions, no: official Turnitin checks are run through the institution's own submission flow, so students cannot simply generate an official AI writing report on a personal draft [4]. Access is controlled by the school, not by the student, which means the first time many students see their AI signals is after the assignment has already been submitted [4].
That gap is why pre-submission checking exists. Independent services such as turnitin0.com let a student upload a .docx, .pdf, or .txt file and receive a Turnitin AI detection report and a similarity report together, so the signals can be reviewed while the draft is still editable [4]. The practical value is timing: interpreting a flagged passage days before the deadline allows revision, whereas interpreting it after submission allows only damage control [4].
When you do review your own signals, work passage by passage rather than reacting to the headline number. Read each highlighted segment, ask whether the phrasing is genuinely yours, and revise the sentences that read as generic or over-polished [4]. Reading the signals early and in context is what turns a stressful flag into an ordinary editing task [4].
Reviewing the signals is only useful if you are looking at the same report your instructor will open. turnitin0.com exists for that exact moment — it returns the Turnitin AI detection report and similarity report together, so the signals you review are the signals that matter, before the submission window closes.
※ Turnitin0.com - Actual [Turnitin AI Report](https://www.turnitin0.com/guides/us/can-i-check-turnitin-for-free-explain-how-schools-use-turnitin) Cover, Score, Flag And Similarity Summary
FAQ
Does an AI signal mean I cheated?
No. A signal is a confidence estimate produced by a detection model, and Turnitin presents the AI percentage as something for a human to interpret rather than a conclusion [1]. Context, citations, and drafting history are what actually determine authorship.
Why does my report show *% instead of a number?
Turnitin displays *% when the AI score falls below its 20% confidence threshold, meaning only low-confidence signals were found [3]. It is not the same as a confirmed 0% and it is not a confirmation of AI use.
Can I see my AI signals before I submit?
Not through official institutional Turnitin, since report access is controlled by your school [4]. Pre-submission checking services return an AI detection report and a similarity report on your own draft so you can review the signals while it is still editable [4].
Are AI signals ever wrong?
Yes. False positives are a recognised limitation and appear most often in formulaic or heavily rewritten academic prose [3]. That is why signals should always be reviewed passage by passage in context rather than treated as proof [3].
What should I do if a passage is flagged?
Read the segment, check whether the wording genuinely reflects your own reasoning and citations, and revise anything that reads as generic [4]. Reviewing early is the whole point of an AI signals review.