How our checks work

The short answer

Plagiarism Checker Plus runs two different checks. The plagiarism check searches web pages and published sources for wording that matches yours and shows each match next to its source. The AI detector estimates, sentence by sentence, how likely each sentence is to have been written by an AI model such as ChatGPT, Claude or Gemini. You can run either one or both on the same text and get one report.

Both results are there to help you review a text. A match is not proof of plagiarism, and an AI score is a likelihood, not proof of who wrote something.

How the plagiarism check works

1. Your text is sent for matching. It goes over an encrypted connection (HTTPS) to our plagiarism-matching service, which compares it with web pages and published and academic sources it can search.
2. Matching passages are found. Wording that is the same as, or very close to, a source is marked, and each match is linked to the source it came from.
3. You get a report. Matches are highlighted in your text, coloured by source, with a list of sources and a percentage: the share of your text that matches sources.

What the plagiarism check cannot do:

• It does not know whether a match is cited. A correctly quoted and cited passage still matches, so it still shows up.
• It does not judge intent. Common phrases, names and set expressions can match without anything being wrong.
• It can only find sources it can search. Text that is not online, such as other students’ unpublished papers, many printed books and closed databases, may not be found.

For a step-by-step method, read how to check for plagiarism and how plagiarism checkers work.

How the AI detector works

The AI detector does not search the web. It looks at how the text is written. AI-generated text tends to be predictable: it keeps choosing the most likely next word, holds a steady sentence length and rarely changes tone. The detector gives each sentence a likelihood of being AI-written and combines them into an overall AI score for the text.

• High: 80% or more. The sentence reads very much like AI-generated text.
• Medium-high: 60–79%. Clear signs, but not conclusive.
• Medium: 40–59%. Could be a person or an AI. From here a sentence counts as flagged.
• Low: under 40%. Most likely written by a person.

AI detection needs at least 50 words. Text you check for AI is sent to an AI-detection service through its API. Under that service’s terms, text sent through its API is not stored by default and is not used to train its models, and we have not turned storage on.

When to take extra care with an AI score:

• Short text. A few sentences give the detector little to work with.
• Formal or formulaic writing. Lab reports, legal text and template-style introductions can read as machine-like even when a person wrote them.
• Non-native English writers. Simpler, more predictable phrasing can be flagged more often.
• Edited text. AI drafts rewritten by a person, or human drafts polished by a tool, mix both styles.

If a detector flagged writing you did yourself, read why AI detector false positives happen.

How to read a result

Look at where the matches or flagged sentences are, not only at the number. One flagged sentence in a long essay rarely means much. A run of high-likelihood sentences, or a long passage matching one source, is a reason to look closer: check the citations, compare with the writer’s other work and ask about drafts and sources. If you review someone else’s work, use the result to start a conversation, not as a verdict.

What we do not claim

• We do not claim that any checker, including ours, is right every time.
• We do not offer tools or advice for getting AI text past detectors.
• We do not use the text you check to train AI models. See privacy and security.

Questions or a result that looks wrong

Tell us through the contact page or at contact@plagiarismcheckerplus.com. Each report also has a feedback button, and we read what comes in.