Is It Possible to Get 0% AI on Turnitin?

A current Turnitin AI writing indicator can show 0%. Since 8 July 2024, zero is the only exact number it surfaces below twenty: results from 1% to 19% are rendered as an asterisk, with no percentage and no highlights. Reports generated before that display change are the exception; Turnitin says they may still show a numerical score below 20%. The indicator is not visible to students either. So the thing people are chasing when they say "0% AI" — a low number they can repeatedly watch, aim at and confirm — is not a state they normally control or observe. It appears on somebody else's screen, under a display rule whose date matters.

The rest of this is what those four are, why the space between zero and twenty has no gradations in it, and what a 0% would and would not settle if somebody did show you one.

Turnitin's current FAQ has a short section headed "What do the different indicators mean?", which lists what appears next to a submission once processing finishes. It runs to four states, and the differences between them matter more than the differences between two numbers.

The bottom two rows are not results at all: in one Turnitin declines to process the file, in the other it tries and fails. Read the top two together and the shape of the current scale becomes obvious. At twenty and above, the indicator behaves like a measurement: a number, and highlighted passages you can go and look at. Below twenty it behaves like a light switch. Zero is one position. The 1% to 19% band is the other.

A report generated under the rule in place since 8 July 2024 will not surface 4%, 11% or 19%; it shows the asterisk instead. An older report can retain one of those numbers, which is why the generation date belongs beside any score you quote. On a current report, "I want to get my score down to something small" describes an outcome the interface cannot grade for you, and "I want 0%" names the only exact sub-twenty figure it will display.

The rule is published, and so is the reason for it:

"False positives (incorrectly flagging human-written text as AI-generated) are a possibility in AI models. 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."

The boundary removes two kinds of information, and the second gets less attention than it deserves. The number goes, which is the part everyone notices. The highlights go too. Below the threshold, the report gives you no passage to inspect and no sentence-level location for the result.

Follow that through to a revision under the current display rule. Suppose the underlying result moved from 17% to 3%. Both reports show an asterisk. Same symbol, same absence of highlights, same everything. The work happened; the interface has no vocabulary for reporting it. Anyone who claims a current report moved "from 15% to 4%" is supplying two exact figures the report did not surface. A dated report from before 8 July 2024 is the exception, not evidence that the current rule works differently.

None of this is a rule about misconduct. It is a decision about what the vendor puts on a screen; the point at which a school reviews a submission comes from that school's own policy. We went through that in is 20% AI too high.

This makes "can I get to 0%" a question about access before it is a question about difficulty. The FAQ explains why in two sentences, and the second is the one that usually gets dropped:

"The AI writing detection indicator and report are not visible to students. However, with the PDF download feature, instructors can download and share the AI report with students."

Elsewhere on the same page: "only instructors and administrators are able to see the indicator." So the only route by which your own AI result reaches you runs through somebody else deciding to download a PDF and send it. That is a real route — plenty of students have seen their report — but it is not a dial you can check.

Which means the loop most advice implicitly assumes does not exist. Submit, look at the score, revise, look again: the two steps that involve looking are not yours to take. What you get instead is a single verdict, later, possibly relayed, possibly as a screenshot with no date on it. And when a tutor says the indicator "isn't showing", that has its own set of documented causes that have nothing to do with your file. We collected those in why your AI score is missing or inconsistent.

Suppose you get one anyway — an instructor downloads the PDF, hands it over, and it reads 0%. Here is what Turnitin says about the design decision behind that number:

"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."

That is a vendor saying, in its own words, that it would rather miss AI text than flag human text. Three consequences follow, and they run in both directions:

• A 0% does not certify that nothing was AI-assisted. The tool is tuned to under-report. A zero is consistent with a clean paper and it is also consistent with a paper the model did not catch, and the report contains nothing that distinguishes the two.
• A 0% is not a statement about your whole file. The percentage is computed over qualifying prose. A document that is largely lists, numeric tables and a reference section had comparatively little assessed, so its zero covers less ground than it appears to. Full paragraphs sitting inside table cells are a separate case: Turnitin says it processes long-form prose in tables. The denominator problem is the subject of how to read a Turnitin AI writing report.
• A 0% is not evidence you can carry into a conversation. Turnitin is explicit that it does not determine misconduct and provides data for educators to decide with. What actually answers "did you write this" is drafts, sources, notes and your ability to talk about the work. That material is listed in flagged, but you wrote it yourself.

A zero is an output of a classifier, run over part of a document, on one day, under one model version. It is worth having. It settles less than the phrase "0% AI" suggests.

A missing number is not automatically a zero. Sometimes there is no result; sometimes there is a non-zero result whose exact number is withheld; and sometimes a short or prose-light file produces a real result over a narrow denominator. Those cases need different names.

• Grey double dash or processing error: no result. Fewer than 300 words of long-form prose, more than 30,000 words, an unsupported language or file type can make the file ineligible; a separate error state means Turnitin failed while processing it. Neither is a clean score, and neither should be reported as 0%. Full detail in can Turnitin check a whole thesis.
• Asterisk: a result above zero whose number is hidden. On reports generated under the current rule, it means the detected range is 1% to 19%. The percentage and highlights are withheld because of false-positive risk. It is not blank, even though it gives you no location to work from.
• Short or prose-light document: still potentially a real result. Non-prose material narrows the denominator; it does not by itself remove the score. If at least 300 words of long-form prose remain, the file can still generate a report. In a document of only a few hundred words, Turnitin says the prediction is mostly "all or nothing" because there is one segment and no overlap to average. That is a warning about extreme results, not a third blank state. The source does not rule out the zero side, and zero sits inside Turnitin's published successful range.

The practical test is simple. A double dash or error means there is no score. An asterisk means there is a non-zero score that is not being shown. A blue 0% is an actual classifier output, even when the file was short or the qualifying denominator was narrow. The three should not be collapsed into one idea of "nothing detected".

Turn all of this around and something slightly odd falls out. The reports that alarm people — the ones at twenty and above — are the only reports that come with enough information to act on. They have a figure and highlighted passages. In the current 1% to 19% band you have neither; at 0% you have the figure but no highlighted passage to revise.

HumanPen is built on that side of the line. Upload the DOCX together with the Turnitin or iThenticate AI report and the highlighted passages become the scope. Only that text is rewritten; everything outside it stays untouched. Charges are based on the rewritten word count rather than the length of the uploaded file.

HumanPen also offers free follow-up support: eligible results can continue lowering AI for free. That support does not promise a particular detector result.

None of that is a forecast. Our own FAQ on that page says it plainly: "Detectors keep changing, so no one can honestly promise a fixed outcome." The percentage on a recheck is unknown; quoting one in advance would be fabrication. Turnitin has never said how a given edit should move a score, and the display rule above means that below the threshold nobody could check such a claim anyway. What we can tell you is which words get touched and which get billed.

Can a Turnitin AI writing report actually show 0%? Yes. Turnitin's FAQ describes the successful state as a blue indicator with a percentage between 0 and 100, and zero sits inside that range. The date caveat concerns 1% to 19%: older reports may retain those numbers, while current ones show an asterisk.

Is an asterisk the same as 0%? No. On a current report, the asterisk means AI was detected in the 1% to 19% range, and Turnitin withholds the number and the highlights in that band because of false positives. It is a non-zero result that the interface declines to quantify.

Can I check my own AI score before my tutor does? Not through the indicator. Turnitin states that the AI writing detection indicator and report are not visible to students, and that instructors can download the report as a PDF and share it. Anything you see, you see because someone chose to show you.

If my report says 0%, am I safe? It is a good result and it is not a certificate. Turnitin says it accepts missing some AI writing in order to keep its false positive rate low, and it says the AI score does not determine misconduct. Both halves of that matter.

Will rewriting get me to 0%? The percentage on a future check cannot be promised. A tool that names one is quoting a result it does not possess. What a report-guided rewrite decides is which passages get changed and which are left exactly as you wrote them.