Source Match Checker

Text to check Our caching layer stores rendered pages for five minutes. When a page changes, the editor service publishes an invalidation event to every edge node, which drops the stale copy before the next request arrives. We chose this design because it keeps reads fast. Sources to compare againstreference answers, earlier submissions, the pages a draft was researched from + Add source Check for copied passages

Using this Tool

Paste the text you want to check, then paste each document it might have been copied from: a reference answer, an earlier submission, the article a draft was researched from, your own published copy. The source match checker highlights every run of eight or more words the text shares word for word with a source, and reports what share of the text's words sit inside those passages. Case and punctuation are ignored, so a passage still matches after its commas are edited.

What it doesn't do

• It doesn't search the web. Only the sources you paste are checked, so the result depends on what you compare against.
• It doesn't detect paraphrase. A passage reworded so that no run of eight words survives isn't reported, and a low score doesn't show a text is original.
• It doesn't judge intent. Quotations, shared boilerplate and a question repeated in its answer all match. Tick Ignore quoted passages to leave out text inside double quotation marks, and read the highlighted passages before acting on the percentage.
• It is tuned for languages written with spaces between words. Chinese, Japanese and Thai text matches poorly.

Raise Shortest passage to report only longer copied runs, or lower it to 6 for short texts such as an answer to a single question. To check whether a text was written by AI instead, use the AI content detector; the two answer different questions and work well side by side.

Need this in a grading, hiring or editorial workflow? The Source Match API takes the same inputs and returns each passage as character offsets into your text. It is free of charge, and no AI model is involved, so the same inputs always give the same result.