Direct answer
An originality checker for an essay compares your draft against a large database of student papers, websites and academic publications, then returns a similarity index plus, in modern systems like Turnitin, a separate AI-writing score [1]. The percentage is a measure of matched text, not a verdict on plagiarism: a high score can come entirely from quotations and reference lists, and a low score does not prove the work is original [1]. What matters is which passages matched, which sources they matched, and whether your citations are doing their job.
What does an originality checker actually check in an essay — and what does the report show?
An originality checker runs your essay through a comparison index and reports back on two different questions that students routinely conflate. The first is textual overlap: how much of your wording also appears in the repository of previously submitted student papers, published articles and indexed web pages [2]. The second, in systems that now include it, is authorship style: whether the prose patterns look machine-generated rather than written by a person [2]. These are separate reports with separate scores, and a clean result on one tells you nothing about the other.
The similarity report itself is more granular than the headline number. Turnitin segments the submission and highlights each matching passage, then lists every source that contributed to the total, so an instructor can see that 18% might be 12% from your own reference list, 4% from a quoted definition and 2% from a phrase that appears in a thousand essays [1]. This is why the guidance on interpreting the report stresses reading the matches rather than the percentage — the number is a summary of a conversation, not the conclusion of one [1].
The AI writing report follows the same logic but a different mechanism. It estimates what proportion of qualifying text appears AI-generated and highlights the specific sentences that drove that estimate, which means you can inspect the flagged passages yourself instead of guessing [2]. Turnitin is explicit that this is an indicator requiring human interpretation and that false positives occur, particularly with heavily edited or formulaic academic prose [2]. For a student, that is both reassuring and a warning: a flagged sentence is a prompt to look closer, not a confession.
Are free online originality checkers and "Turnitin checkers" reliable for a university essay?
The short answer is that no checker outside your institution's own account produces the artefact your instructor will see. Turnitin's student guidance points students to the account or learning-management system their university provides, because students cannot self-enrol into an institutional Turnitin account, and sites advertising themselves as "Turnitin" without that institutional link are not official [3]. That matters practically: a report is only meaningful relative to the comparison repository behind it, and a checker with a smaller or differently maintained database can return a materially different percentage for the exact same essay [3].
There is also a data-handling dimension that rarely appears in the marketing copy. When you upload a draft to an unfamiliar checker, you are handing over unpublished academic work and, in many cases, agreeing to terms that permit it to be retained or added to a database [3]. Some services also differ in whether your submission becomes part of the corpus future students are compared against — which is precisely the mechanism that generates "self-plagiarism" matches later in your degree [1]. Reading the retention terms before uploading is not paranoia; it is the same due diligence you would apply to any service holding your work.
What free tools can legitimately do is give you a rough signal. A free checker that flags a large block of unquoted text is probably right that the block is unquoted, and that is worth fixing regardless of the percentage. What it cannot do is tell you what your university's system will report, because the repository, the exclusion settings and the AI-detection model are all institution-specific [3]. Treat free results as a smoke test and institutional results as the measurement.
If my essay's originality or AI score comes back high, what should I do before submitting?
Start with the source list, not the score. A large share of inflated similarity in student essays comes from correctly handled material — block quotations, reference lists, standard methodological phrasing and discipline-specific terminology — and Turnitin's own guidance on reducing similarity begins by checking whether the matches are actually problems [4]. If a match is a properly cited quotation, the fix is formatting and exclusion settings, not rewriting your argument. If a match is an uncited idea, the fix is a citation. Only the third category, genuinely borrowed wording presented as your own, requires you to rewrite.
When rewriting is needed, do it substantively. Paraphrasing means restating the idea in your own analytical voice and then citing the source of the idea, not swapping synonyms into the original sentence order — synonym substitution leaves the structure intact and produces prose that reads as both derivative and machine-like [4]. The same principle applies to AI-flagged passages: the durable remedy is revising the sentences in your own voice, tightening your own reasoning and keeping drafts that show your process, rather than running text through a thesaurus or a paraphrase spinner [4].
Finally, separate the two scores in your head before you panic. A high similarity score is usually a citation and formatting problem with a clear remedy; a high AI score is a stylistic signal that may or may not reflect how the text was produced [2]. Both are worth acting on before submission, and both are far easier to act on when you can see the flagged passages yourself rather than reasoning from a single number [1]. That is the practical case for checking your own draft early, while you still have time to revise rather than explain.
The gap between "my checker said 9%" and "my instructor's report says 31%" is almost always the repository, not the essay. If you want to see the report your university's system will actually generate — the highlighted matches, the source list and the AI writing score on your own draft — turnitin0 runs your essay through a real Turnitin check and returns both PDFs together, typically in under 15 minutes.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is a high similarity score the same as plagiarism?
No. The similarity index measures matched text, and matches frequently come from quotations, reference lists and common academic phrasing [1]. Plagiarism is a judgement about attribution and intent, which is why instructors read the source list rather than the percentage [1].
Does an originality checker also detect AI-written text?
Not necessarily. Similarity checking and AI writing detection are separate functions with separate reports, and a tool can offer one without the other [2]. Turnitin provides both, and they should be read independently [2].
Can I check my essay without adding it to a database?
It depends entirely on the service. Institutional submissions are typically retained in the comparison repository, while some independent checkers run non-repository checks that do not add your file to the student paper database [3]. Confirm the retention terms before uploading [3].
Will paraphrasing lower my similarity score?
It can, but only if the paraphrase is substantive — restating the idea in your own analytical voice and citing the source [4]. Synonym substitution leaves the original structure intact and often produces prose that reads as both derivative and machine-like [4].
Why did my score differ between two checkers?
Because similarity is calculated against a specific repository, and repositories differ in size and content [3]. The only figure that predicts your institutional result is one produced against the same database your university uses [3].