Plagiarism Checker

You paste a text or upload a document, and the checker selects its most distinctive passages — the sentences long enough, specific enough and rare enough that finding them online would mean something. Each one is searched on the web, in quotation marks, as an exact phrase. The results that come back are compared with your passages, and the ones that genuinely reproduce your text are shown to you with their source: the page, the matching excerpt, and a link.

The result is a percentage of checked passages found online, a highlighted view of your text showing where each match sits, and a list of sources you can open. It is a real search, not an opinion: every source shown exists, every excerpt was actually retrieved, and a match is only kept if the two texts share a run of at least six consecutive words. That last rule is what separates a copied sentence from two texts that merely discuss the same subject.

Three hundred characters are needed at minimum; up to twelve thousand are handled in one check. PDF, Word and text files are accepted. No sign-up is required.

• the distinctive passages of your text are searched on the web, verbatim
• every source shown is a real page, with a clickable link
• matches are verified mechanically: six shared words in a row at least
• PDF, Word and text accepted; no sign-up

At the top, a headline and a percentage: the share of checked passages that were found online. Zero means none of the passages searched turned up anywhere on the open web. Below it, three figures — passages analysed, web searches run, matches found — so you know exactly how much was checked.

On the left, your text with each found passage highlighted; on the right, the sources. Each source card shows the verdict — identical text, or very close — the site it was found on, the passage of yours it matches, a one-sentence note on what was observed, and a link to the page. Click a card and the matching passage lights up in your text.

Under the sources, a fold-out lists the passages that were checked and found nowhere. That list is as much a result as the matches: it is what the tool verified and cleared.

0% — no passage found. The distinctive passages of your text were searched on the web and none was found. For a text you wrote yourself, this is the expected result and the one that lets you submit with confidence. It covers the open web — the sites a search engine can reach — which is where the great majority of copied text comes from.

1 to 34% — a few passages found. One or two of your passages exist online. Open the sources: a match is very often a quotation you cited correctly, a definition, a title, a standard formula or a passage from your own earlier work. If it is a passage you did not intend to copy, the source is right there, and the section below explains what to do.

Over 34% — several passages found. A substantial part of the text exists elsewhere. If the text is yours, check whether it was published before — a blog, a previous submission, a shared document — because your own text being online is the most common cause. If it is not yours, you now have the sources.

• 0%: nothing found — the expected result for your own writing
• a few matches: usually quotations, definitions or your own earlier text
• many matches: the text exists elsewhere — check whether it is yours

Run the check when the text is finished but before it is submitted — that is the moment when a found passage can still be fixed by adding a citation or rewriting a sentence. Paste the body of the work, or upload the document, and go through the sources one by one.

Most matches on an honest piece of work are quotations, and a quotation that is cited is not plagiarism. What the check catches is the other kind: the sentence you copied into your notes months ago and forgot was not yours, the definition lifted from a textbook, the passage paraphrased so lightly that six words in a row survived. Those are the ones to fix, and each one takes a minute once you know where it is.

For a long thesis, check chapter by chapter: twelve thousand characters is about four pages, and a chapter-by-chapter check gives you a finer view of where the matches are anyway. Keep your drafts and notes — if a question ever arises, they are your best evidence.

Open the source and read it next to your passage. Then decide which of four cases you are in. It is a quotation you cited: nothing to do. It is a quotation you forgot to cite: add the reference and, if the passage is long, quotation marks. It is an idea from a source that you paraphrased too closely: rewrite it in your own words — the paraphrasing tool on this site does exactly that, and rechecking afterwards confirms the run of shared words is gone — and cite the source, because a paraphrased idea still needs its author.

It is your own earlier text: a blog post, a previous assignment, a shared document. That is not plagiarism in the everyday sense, but many institutions treat reusing your own submitted work as self-plagiarism, so check your institution's rules before leaving it as is.

In every case the source card tells you exactly which passage is concerned, and the highlight shows where it sits in your text. Nothing has to be hunted for.

People searching for a plagiarism checker are increasingly asking a second question at the same time: was this written by an AI? The two questions are different, and this site answers both with two different tools. The plagiarism checker asks whether a text exists elsewhere already, and looks for it. The AI detector asks whether a text looks generated, and analyses its style.

A text can be fully original and generated: a language model produces new sentences, and a plagiarism checker will find nothing. A text can be copied word for word from a human author and score as entirely human on a detector. So for a complete picture of a text you are unsure about — a submission, a guest article, a job application — run both. They take a minute each, and together they cover what either one alone would miss.

Plagiarism is presenting someone else's work as your own — their words, their ideas, their structure, their data — without attribution. It is an ethical rule before it is a legal one: copyright law protects the expression of a text, but academic and professional honesty protects the idea too, and a paraphrased idea without a citation is plagiarism even where it is not copyright infringement.

Three points follow that people often get wrong. Copying is plagiarism whether or not it is caught. Rewording a source without citing it is plagiarism, however different the words. And quoting a source with a citation is not plagiarism, however long the quotation — it may be poor writing, but it is honest. A checker finds copied words; only you can supply the honesty.

The most common form of plagiarism in student work is not copying but close paraphrase: taking a source's sentence, changing a few words, and presenting it without attribution. It is often unintentional — the writer read the source, took notes in the source's words, and wrote from the notes. It is still plagiarism, and it is what a plagiarism checker catches when six words in a row survive the rewording.

The cure is simple and takes discipline: read the source, close it, write the idea from memory in your own words, and cite it. A paraphrase written that way shares no runs of words with the original. If you find yourself with a passage that is too close, the paraphrasing tool will rebuild it — but the citation is still yours to add.

Every idea, fact or figure that is not common knowledge and not yours gets a reference. Every passage of someone else's words gets quotation marks and a reference. And every source in the references is one you actually used. Everything else — the citation style, the format of the list — is convention, and your institution or publisher will tell you which one.

A text that follows these three rules will produce matches on a plagiarism checker — the quotations — and every one of them will be clean. That is the normal, expected result of honest work. The check is not there to produce zero; it is there to confirm that what is found is what you cited.

• every borrowed idea, fact or figure: a reference
• every borrowed passage of words: quotation marks and a reference
• every reference listed: a source you actually used

Reusing a passage you wrote for an earlier assignment, a blog or a report is not plagiarism in the everyday sense — the words are yours. But many institutions and publishers treat undisclosed reuse of submitted work as self-plagiarism, because a piece of work is supposed to be new. A checker will find your own published text as readily as anyone else's.

The practical rule: if the earlier work is public, or was submitted somewhere, say so and cite it as you would any source. If it is a draft nobody has seen, reuse it freely. When a match points to something you wrote, that is the question to ask.

Not every sentence of a text is worth searching. 'It is important to note that this may vary' exists on ten thousand pages and proves nothing if it is found; a sentence with a date, a name, a figure and an unusual word is found only where it was copied. The checker scores each sentence for how distinctive it is — length, rare words, digits, capitalised names, lexical variety — and discards stock phrases entirely.

It then picks up to eight passages, spread across the whole length of the text rather than clustered in one dense paragraph, so that a copied passage at the end of a document is as likely to be checked as one at the start. Eight is a deliberate ceiling: each passage is a real web search, and a bounded number of searches is what keeps the tool free.

You can see exactly which passages were checked: the found ones are highlighted, and the cleared ones are listed in the fold-out under the sources.

The open web is precisely the right terrain for a second kind of check: content written for publication. An editor receiving a guest post, a marketing team commissioning copy, a site owner wondering whether a page has been scraped, a journalist checking a press release — for all of these, 'does this text already exist online' is exactly the question, and the answer comes with the URL.

Paste the article and read the sources. A match with a competitor's page tells you the copy was lifted; a match with your own site tells you it was scraped; a match with a press release tells you where the 'exclusive' came from. For content you are about to publish, a clean result means it will not be flagged as duplicate by anyone else running the same check.

The plagiarism software used by universities — Turnitin, Compilatio and their peers — reports a similarity percentage: the share of the text that matches something in their databases. That number is routinely misread. A 20% similarity on an essay with many properly cited quotations is normal; a 5% similarity made of one uncited paragraph is a problem. The percentage measures overlap, not honesty.

This checker reports a percentage too, but of a different thing: the share of the distinctive passages it searched that were found online. Read it the same way — as a pointer to the sources, which are what actually tell you whether there is an issue. In both systems, the number is the beginning of the reading, not the end.

Institutional software compares a submission against the open web plus closed collections: previous student submissions, subscription journal databases, essay-mill archives. You do not have access to it until you submit, and by then the result is the marker's, not yours. This checker searches the open web, which is where most copied text comes from, and you can run it whenever you like — while writing, when a citation can still be added.

The two are complementary. Use this checker during writing and before submission, to find and fix what the open web can show you. Expect the institutional check at submission to see the same open-web matches plus anything in its closed collections. A text that comes through this checker clean, with its quotations cited, comes through the other one with only those quotations flagged — which is what an honest text looks like in either system.

It happens: a similarity report flags a properly cited quotation, a standard definition, a passage of your own earlier work, or a coincidence of common phrasing. If you are on the receiving end, the response is calm and factual. Ask to see the report and the specific passages. For each one, show the citation, the quotation marks, or the earlier work that is yours.

Your drafts, notes and document history are your strongest evidence — a text with a history was written, not pasted. Running this checker on the text yourself, and going through its sources, prepares you for exactly that conversation: you will know before anyone else what matches and why.

The text you check is sent to the server, its passages are searched, and the result is returned. It is not stored, not added to any collection, and not compared with anyone else's text. Several free checkers keep every document submitted to them and add it to the database they compare against — which means a thesis checked today can be flagged as 'already existing' tomorrow, by the same service, against itself. This one does not.

The only records kept are a usage counter for the daily limits, and a short-lived cache keyed on a hash of the text, so that checking the same passage twice does not run the same searches twice. The cache expires after a week and cannot be searched or read back as text.

Each check runs up to eight real web searches, and those cost money, so the free allowance is smaller than for the other tools on this site: two checks a day without an account, eight with a free account, and a site-wide daily ceiling that protects the service from a surge. If the ceiling is reached, the tool says so and asks you to come back tomorrow; it never returns a partial or made-up result.

That last point is the rule the whole tool is built on. If the search service is unavailable, the checker refuses to answer rather than say 'nothing found' about a text it did not search. A result on this page always means a search took place.

Picture a twelve-hundred-word essay on the causes of the First World War. Eleven hundred words are the student's own; one paragraph on the alliance system was pasted from an encyclopaedia months ago, into notes, and made it into the essay unnoticed. The checker selects eight distinctive passages spread across the essay — a sentence with a date here, one with two names there — and searches each.

Seven come back with nothing: the cleared list. One, from the alliance paragraph, comes back with the encyclopaedia page: verdict 'identical', a run of twenty-two shared words, the URL. The result reads 13% — one of eight — with the headline 'A few passages found online', the paragraph highlighted in the text, and a source card with the link. The student opens the card, sees the sentence, recognises it, and either rewrites it in their own words and adds the reference, or quotes it with quotation marks and adds the reference. Two minutes. Rechecking gives 0%.

Now picture the same essay entirely the student's own. Eight passages searched, eight cleared, 0%, 'No passage found online', the list of the eight sentences that were checked. That result is the point of the tool: not a verdict of innocence, but a specific, verifiable statement — these eight distinctive sentences of your text do not exist on the open web — that you can submit on.

Verbatim plagiarism is copying words without quotation marks or attribution — what a checker finds most easily. Mosaic plagiarism stitches phrases from several sources into new sentences; the six-word rule catches the phrases that survive. Paraphrase plagiarism rewords a source's idea without citing it; the checker catches it when the rewording stayed close, and only honesty catches it otherwise. Self-plagiarism reuses your own submitted work without disclosure. Accidental plagiarism is any of the above done without intent — notes taken in the source's words, a citation forgotten — and it is treated the same way by most institutions, which is why checking before submission matters.

Source-based plagiarism cites a source incorrectly or cites one that does not say what is claimed; no checker finds it, and it is the reason to keep your sources open while writing. Global plagiarism submits someone else's entire work as your own — a purchased essay, a friend's paper. The open web catches the ones that were published; the institutional check catches the ones that were submitted before.

The most useful thing a teacher can do with a plagiarism checker is let students use it before they submit, and this one is built for that — free, no account needed, instant. A student who has seen a match on their own draft, opened the source and fixed it has learned what plagiarism looks like more thoroughly than any lecture could teach. Make the check a step in the assignment, before the deadline.

For checking submissions, paste the work and read the sources rather than the percentage: a match that is a cited quotation is fine; a match that is not is a conversation. For a suspected case, the source card gives you the passage and the page, which is what a conversation needs. And for the other question — was this generated rather than copied — the AI detector is the tool, with the same advice: read the markers, talk to the student.

• make the check a step of the assignment, before the deadline
• read the sources, not the percentage
• a cited match is fine; an uncited one is a conversation

Outside the classroom, plagiarism is mostly a web problem. Product descriptions copied from a manufacturer, articles scraped from a competitor, blog posts assembled from three others, reviews pasted across sites. Search engines penalise duplicated content, readers notice it, and the original author has rights over it. The checker answers the web question directly: paste the text, and the pages where it already exists come back with their URLs.

For a site owner, that works in both directions. Check a page you are about to publish, and a clean result means it will not be flagged as duplicate. Check a competitor's page that looks familiar, and a match with your own site is the evidence you need for a takedown request. Check a freelancer's delivery before paying for it. In each case the result is not an opinion but a list of pages, and a list of pages is something you can act on.

Run the checker on the finished text. Open every source card and classify it: cited quotation, uncited quotation, close paraphrase, own earlier work. Fix the second and third — add the reference, rewrite in your own words — and decide on the fourth according to your institution's rules. Recheck. Then read the cleared list once, to see which of your sentences were verified, and keep your drafts and notes in a folder in case a question ever arises. Ten minutes, and you submit knowing exactly what any checker will find.

• check the finished text, not the draft
• classify every source card; fix the uncited ones
• recheck, then keep your drafts and notes

• AI detector
• Paraphrasing tool
• Free grammar checker
• Text summarizer
• Free AI chat