Direct answer
A web-based AI checker scans a piece of text and estimates what share of it was likely produced by a generative model such as ChatGPT, Claude, or Gemini. Turnitin's AI writing detection does this by analysing submitted prose and returning a percentage with sentence-level highlighting, and it is built to be read as one signal alongside the similarity report and instructor judgement rather than as a verdict [1]. For a student, the practical value of a web check is seeing the same kind of report your institution sees before the deadline, so a surprise score does not arrive after submission.
What Does a Web-Based AI Checker Actually Detect, and How Does Turnitin's AI Writing Detection Work?
Turnitin's detector examines the statistical texture of submitted writing and reports the proportion of prose it believes was generated by an AI model, then flags the specific sentences that drove that estimate [1]. The output is not a binary "AI" or "human" stamp. It is a percentage attached to a document, with the underlying evidence exposed sentence by sentence so a reader can inspect it.
Inside the report itself, the panel shows an overall percentage together with a breakdown of qualifying words and sentences, and only sentences the model is confident about are highlighted [2]. That distinction matters: the percentage is not a simple count of highlighted sentences, so a document can carry a high score while showing relatively few highlights, or vice versa. Reading the highlights is what tells you where the signal came from.
The detector is also scoped. It targets long-form, prose-style writing in English, which is why a lab report, a reference list, or a heavily bulleted slide deck tends to produce noisy or unhelpful results [1]. If your draft is mostly tables, code, or short answers, a web check is a weak instrument regardless of which tool you use.
Finally, the report reaches instructors through the assignment inbox in the learning management system, which is why the version students preview and the version instructors open are structurally the same document [2]. A web checker is only genuinely useful to a student if it reproduces that document faithfully rather than substituting a proprietary score of its own.
Why Does a Turnitin AI Report Show an Asterisk Instead of a Percentage, and What Does That Mean for My Draft?
When the AI writing score falls below Turnitin's confidence threshold, the report displays an asterisk (*%) in place of an exact figure [3]. The threshold exists because detection on short passages, heavily edited text, or writing that has been restructured sentence by sentence produces unreliable estimates, and publishing a precise number there would overstate what the system knows [3]. The asterisk is therefore a statement about confidence, not a score of zero.
The most common misreading is treating *% as "clean." It is better understood as a low-confidence signal: the system did not find enough consistent evidence to commit to a percentage, which is a different claim from finding no evidence at all [3]. Two drafts can both show *% while one is genuinely human-written and the other is AI-generated prose that has been paraphrased enough to blur the pattern.
For your own workflow, this changes how you should read a check. A high percentage with dense highlighting is a clear prompt to revise. A *% result means the detection is inconclusive, so the sensible next step is to look at the similarity report and at your own drafting history rather than to relax. Because the asterisk is a function of the confidence threshold rather than of your document alone, it is also not comparable across drafts of different lengths [3].
Practically, this is the argument for previewing the real report rather than a lookalike. A third-party tool that returns a single invented number, with no threshold behaviour and no sentence-level evidence, cannot tell you whether you are looking at a confident score or a low-confidence one — and that distinction is the single most useful thing the report gives you [2][3].
What Should I Do If My Draft Is Flagged as AI-Generated Before I Submit It?
First, establish what was actually flagged. Open the report, read the highlighted sentences, and compare them against your own drafts and version history; detection can produce false positives, particularly for non-native English writers and for writing that follows heavily templated structures [4]. If the flagged passages are yours, the evidence of process — earlier drafts, notes, outlines, timestamped files — is what resolves the conversation.
Second, revise rather than argue. Sentences that were drafted with a model and then lightly edited are the ones most likely to retain detectable patterns, so rewriting them in your own voice, adding specific examples from your reading, and breaking up uniform sentence rhythm does more than disputing a percentage [4]. Institutions are advised to treat the report as one piece of evidence among several, which means the burden usually falls on demonstrating authorship rather than on proving the tool wrong [4].
Third, talk to your instructor early if the flag is material to your grade. Turnitin's own guidance frames the report as a starting point for a conversation about how the work was produced, not as an automatic finding of misconduct [1][4]. Raising it before marking is almost always easier than responding to it afterwards.
Fourth, re-check after revising. Because the score responds to sentence-level patterns, a targeted rewrite should move it, and re-running the check confirms whether your edits actually changed the signal or merely reshuffled the same prose [2]. Keep both reports so you can show the trajectory if you are ever asked.
If you would rather see that report before your instructor does, turnitin0 produces the actual Turnitin AI detection report and similarity report from your own .docx, .pdf, or .txt file — the same documents that appear in an institutional inbox, delivered in under 15 minutes in 98% of cases, without adding your paper to Turnitin's student database.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
Get Real [Turnitin](https://www.turnitin0.com/) AI & Similarity Report
FAQ
Is a web-based AI checker the same as Turnitin?
Not necessarily. Many browser tools return their own proprietary score with no sentence-level evidence. Turnitin's report shows a percentage plus the specific flagged sentences, and it is the document instructors actually open in the LMS [2].
Can I check my draft without it being stored in Turnitin's student paper database?
Yes — that is the point of a non-repository pre-submission check. turnitin0 checks your file without adding it to Turnitin's student paper database and does not share reports with third-party databases.
Does an asterisk score mean my paper is clean?
No. *% means the score fell below the confidence threshold, so the system did not commit to a percentage — a low-confidence signal rather than an absence of evidence [3].
How reliable is AI detection overall?
It is a statistical estimate, not proof. False positives occur, especially for non-native English writers and templated writing, so the report is meant to be weighed alongside other evidence [4].
What file types can I check?
turnitin0 accepts English .docx, .pdf, and .txt files between 300 and 30,000 words, under 20 MB.