Paste your text to scan for matching content across the web. Get a similarity score and see exactly which passages match, color-coded by strength.
Paste text or load the example to scan for matches.
See which passages match existing content and where, with each match's strength as a percentage.
Strong matches in red, partial in amber — so you can rewrite the riskiest passages first.
Downloadable .docx reports, deep web-crawl scanning and 30-day history — on Pro.
Plagiarism isn't a single thing, and that trips people up. A teacher worried about a copied essay, an editor checking a freelance draft, and a marketer making sure a blog post isn't lifted from a competitor are all looking for the same underlying signal: text that already exists somewhere else, presented as if it were new. This checker measures exactly that — overlap between what you paste and content that's already published online — and reports it as a similarity score.
What it can't do is read your intent. A high score doesn't prove dishonesty, and a low one doesn't certify originality of ideas. Two writers can cover the same fact in nearly identical wording by coincidence; a properly quoted and cited passage will still light up as a match because the words genuinely appear elsewhere. So treat the result as a map of where your text overlaps with the web — a starting point for judgment, not a verdict handed down. The value is in seeing which sentences overlap and how strongly, then deciding what to do about each one.
When you hit check, your text is split into overlapping passages — short runs of consecutive words rather than the whole document at once. Each passage is compared against indexed web pages and academic content, and the tool looks for sequences that line up closely with something already published. Working at the passage level is what lets a clean paragraph sit right next to a copied one and have only the copied part flagged, instead of smearing one average score across the whole piece.
Matching isn't limited to word-for-word duplication. Light paraphrasing — swapping a few synonyms, reordering a clause — usually still registers, because the structure and the rare, specific terms survive the edit and that's most of the fingerprint. The deeper the rewrite, the weaker the signal becomes, which is honestly why a similarity score and an AI-writing score answer different questions. Each flagged passage comes back with the likely source and a percentage, so a near-verbatim lift reads very differently from a sentence that merely shares a common phrase.
Most people don't run a plagiarism check out of suspicion — they run it because the cost of being wrong is high and a thirty-second scan is cheap insurance:
The number at the top is the share of your text that overlaps with existing sources, and context decides what's acceptable. As a rough rule, under 10% is usually fine — that's the floor you get from quotes, names, citations and unavoidable common phrasing. Somewhere in the 10–25% band is worth a closer look, and anything climbing past 25% generally needs rewriting or proper attribution. These are guidelines, not law; your school, publication or client may set a stricter bar, and theirs wins.
Don't stop at the headline figure — the color-coded passages are where the real work is. Red marks the strongest, near-verbatim matches; amber marks partial overlap. Fix the red first, since one heavily copied paragraph can drag a score up faster than a dozen scattered phrases. A common mistake is panicking over a score inflated entirely by correctly quoted material, or trusting a low score on a very short snippet where there simply isn't enough text to judge. And remember the limits: this scan covers indexed, public content, so a match against a paywalled journal, an unpublished classmate's paper or a private document won't appear. A clean result here means "no public overlap found," which is reassuring — not a guarantee of nothing, anywhere.
Plagiarism-free doesn't mean human. Run the same text through the AI Detector.
Check for AI