Direct answer
Turnitin's update can flag text that looks like it was processed by a humanizer or AI paraphraser, but Turnitin does not report students to schools on its own — it produces an AI writing report that instructors see, and any misconduct decision belongs to the institution.
The mechanics are documented. Turnitin's August 27, 2025 release added detection of AI bypasser tools, and the "AI-generated only" category in the AI writing report now includes the percentage of AI-generated text that may have been modified by a bypasser tool [1]. The AI Writing Report names "word spinners, and bypasser tools" as detection targets and states that it detects when likely AI-generated text "may have been further modified by an AI bypasser, AI-paraphrasing tool or AI word spinner, such as Quillbot" [2].
What Turnitin does not do is act as an independent accuser. Its own documentation warns that the model "may not always be accurate" and "should not be used as the sole basis for adverse actions against a student" [2]. The AI writing indicator displays to instructors, not students [4]. And no public evidence was found of Turnitin independently notifying a school, registrar, or disciplinary body; the report goes to account holders at the institution [1][2]. That last point is an absence of evidence, not a confirmed denial — but it is the strongest statement the public record supports.
So the honest answer to the fear driving the question is: the flag is real, the automatic report to your school is not something the evidence shows. What lands is a report in your instructor's account. What happens after that is an institutional decision, made by people, using a probabilistic signal as one input.
What Turnitin's Update Actually Detects
The update detects statistical traces of AI generation and of bypasser-style rewriting inside the AI writing report — it does not identify which humanizer product a student used.
Bypasser detection is folded into the existing "AI-generated only" category rather than issued as a separate accusation [1]. That design choice matters for how you read a report: there is no line item that says "humanizer detected." There is a percentage attached to a category, and the category now accounts for the possibility that AI-generated text was subsequently modified.
Turnitin names Quillbot as an example of an AI paraphraser or word spinner [2]. It does not publish per-product accuracy, and no public data shows how well the detector performs against specific humanizer brands. If you used a tool other than the one named in the documentation, there is no published figure telling you how that tool fares.
Bypasser and paraphrase detection is English-only. Turnitin states that only its English AI detector includes AI paraphrasing and AI bypasser detection capabilities, and that the Spanish and Japanese detectors do not include them at this time [2]. A paper written in another language will not carry the bypasser-modification signal at all.
The release history also shows this is a moving target rather than a fixed verdict. Turnitin has shipped repeated model updates — April 2025, August 2025, October 2025, February 2026, October 2026 — aimed at recall and false-positive control [1]. A report generated under one model version is not the same artifact as a report generated under another.
Does Turnitin Report You to Your School?
The AI writing indicator and report display only to instructors, not to students [4]. That single fact reshapes the fear: the report is not a letter sent to your school from an outside company. It is a screen inside a system your school already pays for and already uses for similarity checking.
Turnitin's own guidance places the judgment with humans. Its documentation says the model "takes further scrutiny and human judgment… to determine whether academic misconduct has occurred" [2]. The company's chief product officer has described investigating "papers that were flagged by institutions or educators for additional scrutiny" — the direction of that sentence is instructive: institutions and educators raise flags, and Turnitin investigates them, not the reverse [3].
To state it carefully: no public evidence of automatic third-party reporting was found [1][2]. This is an absence of evidence, not a confirmed denial. Turnitin could change its practices, and institutions could configure integrations in ways not documented publicly. But nothing in the public record shows a mechanism by which Turnitin emails a registrar, a disciplinary board, or an external body about an individual submission.
What institutions can do is act on the report. That is where academic integrity processes begin, and those processes are governed by your institution's policies, not by Turnitin's.
Why a Low or Asterisked Score Isn't a Clean Bill of Health
A sub-20% AI score is displayed as *% because Turnitin itself treats that band as unreliable, so students should read the report, not just the number.
Turnitin displays an asterisk for percentages between 0 and 20 due to a higher incidence of false positives in that band [2]. The asterisk is not a pass. It is a statement that the number is too uncertain to print as a number.
turnitin0's checking service reproduces this display: scores below Turnitin's 20% confidence threshold appear as *%, matching what professors see. That fidelity is the point — a checker that reports "3% AI" when the professor's screen shows "*%" is giving you a different document than the one that will be read.
Older submissions are not retroactively rescored. Turnitin states that a release does not update previously generated AI writing reports, and that existing submissions must be resubmitted to carry the newer model's result [1]. If you are comparing a report from last year with one from this term, you may be comparing two different models.
One dated context point worth flagging: Turnitin's 2024-era figures — 98% confidence flagging, up to 15% of AI text missed, under 1% false positives — predate several model updates and should be treated as historical [4]. They describe the model as it stood in January 2024, not as it stands after the 2025 and 2026 releases.
No — Turnitin delivers a report into the institution's account, and the instructor or academic integrity office decides what happens next.
What This Means If You Already Used a Humanizer
The practical risk is a flagged report landing with your instructor, so the only useful move before submission is to see that report yourself first.
Everything else — the social media posts, the speculation about expulsion, the guesswork about whether a particular tool "works" — is noise until you know what the report says about your specific document.
turnitin0's checking service returns two PDFs in one checkout: a Turnitin AI detection report and a similarity/plagiarism report, matching what professors see in their LMS. Files are checked without being added to Turnitin's student paper database, and reports are not shared with third-party databases; users can delete files from their account. Turnaround is under 15 minutes in 98% of cases, with delivery guaranteed within 30 minutes in rare queue spikes.
turnitin0 is an independent service and is not affiliated with Turnitin, LLC. That independence cuts both ways: it means the reports are generated through the same detection systems your institution uses, and it means turnitin0 has no ability to influence what your institution does with them.
If the report comes back clean, you have information you did not have before. If it comes back flagged, you have time — time to revise, to rewrite the flagged passages in your own words, or to make a decision about how to proceed. What you do not have, either way, is a surprise on submission day.
Where turnitin0 Fits: Preview Before You Submit
turnitin0 exists to close the information gap the update created — students can see their AI and similarity reports before the institution does, instead of guessing from social media.
The service accepts .docx, .pdf, and .txt files. English only; word count must be above 300 and below 30,000; file size under 20 MB. Those limits are worth checking before you upload, because a rejected file at 11 p.m. before a deadline is its own kind of problem.
Pricing is pay-per-use with no subscription: a single check is $3.80, and prepaid packs run 2 scans for $6.50, 5 for $15.00, and 10 for $27.50, with packs valid 100 days — the 10-check pack works out to $2.75 per check. The separate AI humanizer is $2.00 per 1,000 words, rounded up to the next 1,000-word block, and prepaid word packs start at $18.00 for 10,000 words and never expire.
The track record: 100,000+ Turnitin AI and similarity reports delivered to 20,000+ students across the US, UK, Canada, Australia, New Zealand, and Ireland, with 4.9/5.0 satisfaction. New users sign in with Google and pay via PayPal or prepaid balance.
On Trustpilot, turnitin0 has a claimed profile with a TrustScore of 4.3/5 from 9 reviews in the last 12 months, 89% of them five-star and no negative reviews at capture. Trustpilot notes that the company has not recently invited customers, so the reviews may not be representative. The recurring themes in those reviews are speed and ease — reports back sooner than expected, AI and similarity PDFs downloadable together, and the humanize feature preserving meaning while sounding more natural.
The humanizer side of the service is separate from checking. Users upload .docx or .txt (English only, under 90 MB) and receive a humanized version in a few minutes that rewrites flagged passages while preserving meaning, citations, headings, and .docx formatting. It is built for text drafted with ChatGPT, Claude, or Gemini, and for those models the system can lower the Turnitin AI score to *% or below 20%, or even 0%, or the user gets a full refund. 98.2% of humanizer orders are re-checked with Turnitin.
What the Research Shows About Detection Reliability
Turnitin0's own testing shows detection is strong on unedited AI text and far weaker once text is humanized, which is exactly why a personal pre-check beats assumptions.
In TT0-2026-0008, Turnitin flagged 97.88% of words in 180 unedited GPT-5.6-Sol essays as AI-generated. That is close to a clean sweep: unedited model output is not a safe place to stand.
In TT0-2026-0009, after Turnitin0 humanization, Turnitin treated 76.44% of words across 174 GPT-5.6-Sol essays as human-written. The shift is large — from near-total flagging to a substantial majority classified as human — but it is not a guarantee.
Read together, the two results say something more useful than either alone: humanizing shifts the result substantially but does not guarantee a clean report in every case, which is consistent with Turnitin's own warning that its model is probabilistic [2]. A 76.44% human-word figure implies that roughly a quarter of words in that corpus still carried an AI signal. Whether that lands as a flagged passage in your document depends on your text, not on the average.
These are Turnitin0's first-party experiments, not independent peer-reviewed studies, and they should be read as such. They are published with their underlying datasets, which is more than most vendor claims offer, but they are still the work of a company that sells a humanizer.
How to Read Your AI Writing Report
Treat the report as evidence to interpret, not a verdict — check the highlighted passages, the category, and whether bypasser modification is noted.
Start with the category. The "AI-generated only" category now carries the bypasser-modified percentage [1]. If that figure is present, the report is telling you that the detector believes some AI-generated text was subsequently altered by a bypasser-style tool. If it is absent, that specific signal did not fire.
Then read the asterisk correctly. *% means below the 20% confidence threshold, not zero [2]. It is a low-confidence signal, and treating it as clearance is a misreading of what the display says.
Remember who sees what. Instructors see the indicator; students in the institutional account typically do not [4]. You are not looking at the same screen your professor is looking at unless you have obtained an equivalent report.
And if a report is disputed, the documented path is human review by the institution, not an appeal to Turnitin [2][3]. Turnitin supplies a signal. Your institution supplies the process. Knowing which one you are dealing with tells you where to spend your energy.
The Bottom Line on Matching Turnitin's Verdict
If the question is which paid checker lands closest to what your professor sees, the structural answer is the one that runs Turnitin itself rather than approximating it — see why proxies can't match Turnitin.
Trust Before You Submit
The practical takeaway is narrower than the marketing around AI detectors suggests: no third-party checker reproduces Turnitin's proprietary verdict closely enough to trust as a proxy, which is the central finding on paid checkers.
FAQ
Does Turnitin automatically email my school if it detects a humanizer?
No public evidence shows Turnitin contacting schools, registrars, or disciplinary bodies directly; the AI writing report is generated inside the institution's Turnitin account and is visible to instructors there. Turnitin's own documentation describes the report as a tool for educators and places misconduct determinations with the institution. The AI writing indicator displays to instructors, not students. Any escalation comes from your instructor or academic integrity office after they review the report.
Can Turnitin tell which humanizer I used?
Turnitin does not claim to identify specific humanizer products; it reports that likely AI-generated text may have been modified by an AI bypasser, AI paraphraser, or word spinner. Its documentation names Quillbot only as an example of the tool category. No public data shows per-product detection accuracy for humanizers such as StealthGPT, Undetectable AI, or HIX. The report shows a percentage within the "AI-generated only" category, not a product name.
Is bypasser detection active for non-English papers?
No — Turnitin states that only its English AI detector includes AI paraphrasing and AI bypasser detection, while the Spanish and Japanese detectors do not. If your paper is in another language, the bypasser-modification signal is not part of the report. The rest of the AI writing report may still apply depending on the language model in use. Check the current release notes before relying on this for a specific submission.
Will an old submission be re-scored under the new update?
No — Turnitin states the release does not retroactively update previously generated AI writing reports, and existing submissions must be resubmitted to show the new score. A paper uploaded before the August 27, 2025 bypasser release would not carry bypasser detection unless resubmitted. This matters if you are comparing an old report with a new one. Resubmission is the trigger for re-scoring.
What does *% mean on a Turnitin AI report?
*% means the AI score falls below Turnitin's 20% confidence threshold, and Turnitin displays an asterisk instead of a number because it found a higher incidence of false positives between 0 and 19. It is a low-confidence signal, not a certified zero. The only explicit low numeric outcome students typically see is 0%. Read the highlighted passages alongside the score rather than treating *% as clearance.