Docs · Plagiarism
Plagiarism.
A just-in-time check against the live web. We pull distinct phrases from your text, search the open internet for an exact match, and show which lines already exist elsewhere, with the source.
How it works
- Your text is broken into distinct phrases. We sample phrases across the whole piece rather than checking every line, to stay fast and within the search tier.
- Each sampled phrase runs as an exact-match web search.
- A phrase that turns up word-for-word on a public page is flagged, scored for similarity, and the page is listed as a source.
The verdict
- No web matches — none of the sampled phrases turned up verbatim online.
- Some overlap — a few matches; short common phrases can match by chance, so review the sources.
- Heavy overlap — many phrases found verbatim, which reads as copied or closely lifted.
What it can and cannot see. This finds copied passages on the open web, not paraphrase, and it cannot see paywalled or academic databases. Treat the result as a strong signal, not a court verdict, and check the sources before drawing conclusions.
The web check runs on the deployed site and needs a search key configured. It is the one tool that spends an external search budget per run.