An originality report is the document a plagiarism checker generates for a submission: it lists every passage of your text that matches an existing source, shows which source each passage matches, and lets you or your instructor open those sources and compare them side by side. The percentage you see, the similarity or originality score, is just the one-line summary of that document, and the report underneath it is where every real answer lives.
What is an originality report?
It is the evidence file behind the number. When you submit work through Turnitin, SafeAssign or Google Classroom, the system compares your text against its database of web pages, publications and previously submitted student work, then assembles a report of everything that matched. Each match is inspectable: the report highlights the block of your text, names the source it matched, and links to it. That inspectability is the whole point. A percentage on its own cannot tell anyone whether the matched text is a properly quoted paragraph, a shared bibliography, or an uncited copy. The report can, because it shows exactly which words matched what.
The most useful mental model is a distinction between two things that usually get collapsed into one. The score is a volume measurement: how much of the text matched anything at all. The report is the qualitative record: what matched, where, and against which source. Instructors who use these tools well spend their time in the report, and students who want to understand a surprising score should do the same, because the same 27% can be entirely defensible or a genuine problem depending on what the matches turn out to be.
What does an originality report actually contain?
Blackboard's SafeAssign documentation gives the clearest published anatomy, and its four sections generalize well to the other platforms. Its report is divided into a Report Summary, a Citations panel, the Submission itself, and a Document Properties section. The summary carries the overall percentage and a risk level. The Citations panel lists every matched source, color coded, with up to 30 distinct colors for 30 different sources, and each source can be opened in its own window. The Submission panel shows your paper with every matching block highlighted in its source's color, and clicking a block shows the original text next to yours.
The fourth section is the one students rarely know exists. Document Properties exposes file metadata to the instructor: who created the file, when it was created, when it was last modified, how many revisions it went through, and what software produced it. Turnitin's report has its own equivalents of the first three sections, with matched sources ordered by match size and the same open-and-compare mechanics. The practical takeaway is the same on every platform: the report is not a score with decoration, it is a structured case file, and everything in it is designed to be checked rather than trusted.
Is an originality report the same as a similarity report?
Functionally yes, and the different names are the single biggest source of confusion around this term. We checked all three vendors' current documentation on the same day, 31 August 2026, and they genuinely disagree on what to call the same kind of document. Turnitin's guides now use Similarity Report throughout; the older Originality Report name survives mainly in its product tier names, such as Originality Check, and in a decade of syllabi and lecture slides that still say originality report. Blackboard's SafeAssign documentation still officially calls its document an Originality Report. Google Classroom calls its feature originality reports, lowercase, and it is Google's own system rather than a rebadged Turnitin product.
So if your instructor says originality report and your screen says Similarity Report, or the reverse, nothing is out of sync except the vocabulary. The document class is the same everywhere: a matched-source evidence file with a summary percentage on top. Which name you see depends on which product your institution bought, not on any difference in what the system checked.
What does the originality report mean on Canvas?
Canvas does not generate originality reports itself. When you see one there, Canvas is displaying the output of an integrated checking service, almost always Turnitin, inside its own interface. The percentage icon appears in SpeedGrader and in your submission view, and clicking through opens the full Turnitin report. That pass-through design has a documented quirk: the score shown in SpeedGrader can differ from the score shown in the separate Turnitin assignment inbox for the same paper, because the two interfaces do not always reflect filter changes at the same time.
How to read that number, and why the SpeedGrader figure is not automatically the final one, is covered in what the similarity score means on Canvas.
The color bands that Canvas and Turnitin attach to the score, and what each band actually implies, are broken down in Turnitin report colors explained.
What does the originality report mean on Blackboard?
On Blackboard the name is exact: SafeAssign's document is officially an Originality Report, and its summary works differently from Turnitin's color bands. SafeAssign classifies the overall match percentage into three risk levels, and its own documentation publishes the boundaries: Low is under 15%, Medium is 15 to 40%, and High is over 40%. The same documentation is unusually direct about what the number is for, describing the overall score as a warning indicator only and telling instructors to review the paper to see whether matches are properly attributed. A Low score typically reflects quotes and common phrases; a Medium score reflects extensive quoted or paraphrased material; a High score means the paper deserves a close read, not that misconduct has been established.
One SafeAssign behavior is worth knowing because it differs from Turnitin: per its documentation, SafeAssign does not check a new attempt against the same student's earlier attempts on the same assignment, so resubmitting your own revised draft does not match you against yourself there. Turnitin can, which is why its guidance warns that submitting your own earlier draft to an institution's private repository can produce a score near 100%. Same category of tool, materially different behavior at the edges, and both behaviors are published rather than secret.
How is a Google Classroom originality report different?
Google's version is built into Classroom and runs on Google's own index rather than a proprietary essay database. Per Google's current help pages, it checks a submitted Docs, Slides or Word file against Google's index of the open web and its scanned books, marks passages that lack a citation, and, where a school administrator has switched on the school matches option, adds a comparison against work submitted earlier at the same school or district. The limits are concrete and published: on the free Education Fundamentals tier a teacher can enable originality reports on 5 assignments per class, teachers can find a generated report for 45 days, files must be under 2MB, and students get 3 practice runs on their own work before submitting, after which the teacher still receives an automatic report on the final submission.
Those caps, and what Classroom's reports can and cannot catch, are covered in more depth in does Google Classroom detect AI.
Does an originality report detect AI writing?
No, and this is the misunderstanding that causes the most real-world trouble. An originality report is a matching system: it can only flag text that resembles something already in its index. Text generated by ChatGPT or any other model is statistically novel, matches nothing, and typically sails through an originality check with a low score. AI detection is a separate system that estimates whether prose was machine-written from its statistical texture, with no source to inspect and documented false positives. Some platforms show both results in the same interface, which is exactly why people conflate them, but a clean originality report says nothing about AI and a high AI flag says nothing about plagiarism.
The AI side of that split, including how to read Turnitin's separate AI percentage, is covered in what the Turnitin AI score means.
If you want to see what a detector would say about your text before anyone else does, our free AI detector gives you that reading in advance.
And if your genuinely original writing keeps reading as machine-generated, that is the specific problem our AI humanizer is built to fix, which is a separate issue from anything an originality report measures.
How do you read an originality report without panicking?
- Open the full report before reacting to the number. The percentage ranks papers for attention; the matches decide what it means.
- Check each highlighted block against its source. Quoted and cited overlap is scholarship working as intended, not misconduct.
- Look at where matches cluster. A bibliography and quoted passages are expected; matched text inside your own analysis is what deserves a rewrite or a citation.
- Know the platform's frame before comparing numbers: Turnitin summarizes with color bands, SafeAssign with Low, Medium and High risk levels, and Classroom with flagged passages, so the same paper reads differently on each.
- Remember the edge cases the vendors themselves publish: Turnitin notes a resubmission can take a full 24 hours to produce a fresh report, and a very large document can display 0% while the report still contains matches.
How this was made: the platform naming, the SafeAssign report anatomy and its 15% and 40% risk boundaries, the 30-source color limit, the Document Properties metadata list, Google Classroom's 5-assignments-per-class, 3-student-runs, 45-day and 2MB limits, and Turnitin's color bands, 24-hour resubmission delay and 0% rounding behavior were all read directly from Turnitin's, Blackboard's and Google's own live documentation pages on 2026-08-31, in the browser rather than from cached copies or third-party summaries. No threshold in this post is invented, and no claim is made about any institution's local policy, because acceptable percentages are set by instructors and institutions, not by the tools. Drafting is AI-assisted, with a person verifying each claim against its source before publishing.
Frequently asked questions
Is an originality report the same as a plagiarism report?
They are the same document under different informal names. Strictly, an originality report shows text matching, not plagiarism: it flags overlap with existing sources and leaves a human to judge whether that overlap is quoted, cited, coincidental or copied. Turnitin's own student guidance states plainly that it does not check for plagiarism, it highlights similarity to its database.
What is a good originality report score?
There is no universal threshold, and no platform publishes one as policy. SafeAssign's own documentation calls scores under 15% low risk and typical of papers with some quotes and common phrases, while Turnitin's guidance says every school, instructor or assignment can set a different acceptable amount. What the matched text is matters more than the percentage: a well-cited 30% can be fine and an uncited 8% can be a real problem.
Why is my originality report 100%?
The most common cause is matching against your own earlier submission. Turnitin's guidance describes exactly this scenario: submit a draft to an institution's private repository and your final version can come back at or near 100%, because it matches the stored draft. The fix is administrative, asking the instructor to exclude the earlier submission, not a rewrite.
How long does an originality report take to generate?
First submissions usually produce a report within minutes, but Turnitin's documentation warns that an overwritten or resubmitted paper may not generate a new Similarity Report for a full 24 hours, a deliberate delay so the resubmission does not match your own previous draft. Google Classroom's reports run automatically at submission, and teachers can find them for 45 days.
Does an originality report check for AI-generated text?
No. Originality reports match submitted text against indexed sources, and AI-generated text usually matches nothing because it is statistically new. AI writing detection is a separate statistical system, reported separately where it exists at all. A submission can score 0% on originality and still be flagged by an AI detector, and the reverse.
