What does the asterisk (*%) mean on a Turnitin AI score?

The asterisk means Turnitin's detector found AI in the 1% to 19% range, and it is not showing you the number or the highlights. The reason, in Turnitin's own words, is that scores in that band carry a higher risk of false positives. So the report is not saying "nothing detected" and it is not saying "zero". It is saying: something was found, and we are not going to show it to you. That has a specific consequence for anyone trying to read or revise against a result in that range, because the two things you would need to act on a score are both gone. The number is gone, and the highlighted passages are gone.

Here is the full answer from Turnitin's AI writing detection capabilities FAQ:

"To avoid potential incidence of false positives, no score or highlights are attributed for AI detection scores in the 1% to 19% range. When AI is detected below the 20% threshold in the report, it is now indicated with an asterisk (*%) and no percentage is attributed."

Three things are packed into that sentence, and the second one usually gets skipped.

First, the asterisk is not a placeholder for zero. It covers everything from 1% to 19%. A report showing an asterisk has a non-zero result behind it, and the report has decided not to surface which non-zero number.

Second, the reason is false positives. The rule exists because Turnitin believes scores in that band are more likely to be wrong, where "wrong" means flagging human-written text as AI. The sentence two paragraphs before this one in the FAQ uses the phrase "incorrectly flagging human-written text as AI-generated" to describe what a false positive is, and the asterisk rule is the direct answer to that risk.

Third, the threshold is 20%. Twenty is where the report starts behaving like a measurement again. Below twenty it behaves like a switch with one position for zero and one position for everything else. The asterisk is one of four indicator states the report can show; it is the only one that says "we found something and we are not telling you what".

The asterisk removes two kinds of information, and the second gets less attention than it deserves.

The number goes. That part everyone notices. The percentage that would have sat between 1% and 19% is replaced by a symbol, and there is no way to read the real figure off the report.

The highlights go too. On a score at 20% or above, the report marks passages it believes are AI-written, and you can go look at them. On an asterisk, there is nothing to look at. No passage is marked. No sentence-level location is given. The report has a result it is not showing you, and it is also not showing you where that result came from.

Follow that through to a revision. Suppose an earlier version of the document came back at 17% and the current one comes back at 3%. Both reports show an asterisk. Same symbol, same absence of highlights, same everything on screen. The work happened. The interface has no vocabulary for reporting it. Anyone who tells you a current report moved "from 15% to 4%" is supplying two figures the report did not surface. That is not a reading of the report; that is a claim about what is behind it.

This is also why comparing two asterisks is not a comparison. Two reports with the same symbol can have different underlying results, and neither report tells you whether they do. If you need to compare two versions of a document, the reports that can be compared are the ones at 20% and above, where the number and the highlights are both visible. We walked through what a readable comparison looks like in how to read a Turnitin AI writing report.

The asterisk rule is about display. There is a separate characteristic of the model that makes the band behind the asterisk less predictable than the number it replaces would have been, and it sits on the same boundary.

From the FAQ, on short documents:

"In shorter documents where there are only a few hundred words, the prediction will be mostly 'all or nothing' because we're predicting on a single segment without the opportunity to overlap."

The next sentence:

"This means that some text that is a mix of AI-generated and original content could be flagged as entirely AI-generated."

So a short document with mixed content can get a result that looks like one thing, when the underlying situation is another. If the reported score on such a document had been, say, 14%, the real picture might be more complicated than 14% suggested. The asterisk hides that 14%, and in doing so it also hides the fact that the 14% itself was an all-or-nothing prediction on a single segment.

None of this is a statement about misconduct. It is a statement about what the tool can and cannot tell you in the band it declines to show. The consequence for a reader is that an asterisk on a short document is the least informative state the report can be in while still saying something happened. And the one thing the report does say, it says without context.

One more thing the asterisk hides is the size of the denominator behind it, and that denominator is narrower than the file.

Turnitin says it analyses only qualifying text, which the FAQ defines as:

"This qualifying text includes only prose sentences, meaning that we only analyze blocks of text that are written in standard grammatical sentences and do not include other types of writing such as lists, bullet points (short non-sentence structures), or other non-sentence structures."

The next sentence:

"This percentage is not necessarily the percentage of the entire submission."

So a document built mostly from lists, tables and a bibliography left the model with very little prose to score. The asterisk on such a file sits on top of a denominator that was already smaller than the document appears. If the hidden figure was 8%, that 8% was 8% of qualifying prose, and the qualifying prose might have been half the file.

This matters for two reasons. The percentage the report is not showing you is not the percentage you would guess from reading the document. And the highlights the report is not giving you would have been the only way to find out what counted. Without the highlights, the denominator the asterisk was computed over is also invisible. You have a symbol standing for a number standing for a fraction of something you cannot measure from the outside.

The asterisk rule is a display decision, but it sits on top of a deeper design choice that goes in the same direction. Here is Turnitin on that choice:

"In order to maintain this low rate of 1% for false positives, there is a chance that we might miss some AI written text in a document. We're comfortable with that since we do not want to incorrectly highlight human-written text as AI-written. For example, if we identify that 50% of a document is likely written by an AI tool, it could contain as much as 65% AI writing."

Read that next to the asterisk rule. The model leans toward catching less rather than flagging too much. The display rule takes the band where the two errors are hardest to separate, and declines to put a number on it. Those are the same direction. A tool that would rather miss AI text than flag human text is also a tool that would rather show an asterisk than show you a 7% it cannot stand behind.

The false positive rate Turnitin publishes is specific about scope. From the same FAQ:

"We strive to maximize the effectiveness of our detector while keeping our false positive rate - incorrectly identifying fully human-written text as AI-generated - under 1% for documents with over 20% of AI writing."

The next sentence:

"In other words, we might flag a human-written document as AI-written for one out of every 100 fully-human written documents."

The clause at the end of the first sentence is the one that connects to the asterisk. The under-1% target is stated for documents with over 20% of AI writing. Below that threshold the target is not given as a number, and the display rule is what answers it: the report simply does not surface scores it cannot stand behind. The 20% boundary in the asterisk rule and the 20% boundary in the false-positive target are the same line.

An asterisk is a result that is not blank and not a number, and it does not leave you a passage to inspect. That narrows what can be done with it directly, but it does not leave nothing.

• Do not treat it as zero. The report found something. It is withholding the figure, not denying one. If you are being asked about an asterisk, the question is what was found, not whether anything was found.
• Ask for the generation date. Reports are produced under the rules in force when they are generated, and a screenshot of an asterisk with no date tells you nothing about which rule produced it. The same applies to other report states: the file is a snapshot, not a live reading.
• Ask what else is on the report. Turnitin says the AI writing detection indicator and report are not visible to students, and an instructor has to download and share the PDF. The asterisk is one item on a document that also contains the Similarity score, file information, and processing notes. Reading the rest of it can clarify whether what you are looking at is current and which report it is.
• Do not try to compare two asterisks. The reports that can be compared are the ones at 20% and above, where both the number and the highlights are visible. Two asterisks can represent different underlying figures and neither report will tell you whether they do.

If a report comes back at 20% or above, that is the band where the report gives you something to act on: a figure and highlighted passages. A report-guided rewrite imports those passages as its scope, and only those passages are touched. If a later report still flags passages, those can be re-run at no charge. Below 20% there are no highlights to import, so the question does not arise.

None of that is a forecast. Turnitin has never said how a given edit should move a score, and the asterisk rule means below 20% nobody could check such a claim. What a report-guided rewrite decides is which words get touched and which stay exactly as written.

Does the asterisk mean my score is zero? No. The asterisk covers the 1% to 19% range. Turnitin says "no score or highlights are attributed for AI detection scores in the 1% to 19% range" and that scores in that band "are now indicated with an asterisk (*%) and no percentage is attributed". Zero is a separate state with its own indicator. The asterisk means something was found and the number is being withheld.

Why does Turnitin show an asterisk instead of a number? To avoid false positives. Turnitin's FAQ says "To avoid potential incidence of false positives, no score or highlights are attributed for AI detection scores in the 1% to 19% range." Scores in that band carry higher false-positive risk, so the report declines to surface them.

Can I find out the real number behind the asterisk? Not from the report. The number and the highlights are both withheld in that band. There is no way to read the underlying figure off a current report showing an asterisk.

Can I compare two reports that both show an asterisk? Not meaningfully. Two asterisks can have different underlying scores and neither report tells you whether they do. The reports that can be compared are the ones at 20% and above, where the number and the highlights are both visible.

Does the asterisk mean the report is broken? No. It is a display rule, not an error. A separate grey double dash means the file could not be processed, and an exclamation mark means processing failed. The asterisk means processing succeeded and found something, and the result is being withheld.

Is 20% the boundary where the asterisk stops? Yes. At 20% and above the report shows a number and highlighted passages. Below 20%, the only exact number the report surfaces is 0%. Everything between 1% and 19% is the asterisk. We covered what 20% means as a boundary in is 20% AI too high.