Does Turnitin detect copy and paste

Turnitin runs two independent systems. The Similarity Report detects copy and paste. It compares your submitted text against a database of published content, other student papers, and web sources, and it flags matching strings. If you copy-pasted from a source that is in the database, the Similarity Report will show it. The AI Writing Report does not detect copy and paste. It analyzes word probability patterns in prose to estimate whether text was AI-generated. Copy-pasted human-written text from a book or an article does not trigger the AI detector, because human writing has different probability patterns than AI-generated text.

These two scores do not influence each other. We will walk through how each system works and what that means if you pasted text from somewhere.

The Similarity Report is the one that does what most people mean when they ask "does Turnitin detect copy and paste." It takes your submitted document and compares it against Turnitin's collection of content. That collection includes published work, web pages, and other student papers that have been submitted to Turnitin before. When the system finds a string of text in your document that matches a string in the database, it highlights the match and counts it toward the similarity percentage.

The mechanism is string matching. If you pasted a paragraph from a website, a sentence from a book that has been digitized and indexed, or a section from a friend's paper that was previously submitted, the Similarity Report will find the match if the source is in the database. The report shows you exactly which strings matched and which source they came from.

Turnitin's AI writing detection capabilities FAQ states this separation directly:

"The Similarity score and the AI writing detection percentage are completely independent and do not influence each other."

The next sentence on that same page is:

"The Similarity score indicates the percentage of matching-text found in the submitted document when compared to Turnitin's comprehensive collection of content for similarity checking."

That second sentence describes exactly what the Similarity Report does. It finds matching text. It does not assess whether that text was written by a human or by AI. It does not care who wrote it. It cares whether the same string exists somewhere in the database.

The AI Writing Report operates on a completely different principle. It does not compare your text against a database of strings. It runs your text through a model that assigns probability scores to segments of prose. The FAQ describes the mechanism this way:

"When a paper is submitted to Turnitin, sentences from the submission are extracted and segmented into overlapping sections for prediction analysis. Each segment is classified by the AI detection model and given a value between 0 and 1, denoting the probability of the text being likely human or AI-generated. Each qualifying sentence within these segments inherits the segment's score. Since segments overlap, some sentences may have multiple scores, which are then pooled into a single score. These sentence scores are further aggregated and used to compute the overall document AI writing score."

The key phrase is "the probability of the text being likely human or AI-generated." The model is looking at how words are chosen and sequenced. AI-generated text tends to produce words that are highly predictable given the preceding context, because language models pick high-probability next tokens. Human writing tends to be less predictable in its word choices. That difference in probability patterns is what the model reads.

This is why copy-pasted human text does not trigger the AI detector. If you paste a paragraph from a published book into your paper, that paragraph was written by a human. Its word probability patterns look like human writing, not like AI-generated text. The model classifies it as human. The AI Writing Report does not flag it.

The reverse is also true. If you generate a paragraph with an AI tool and paste it into your paper, the AI Writing Report may flag it, because the probability patterns in that paragraph look like AI output. But the Similarity Report will not flag it, because that exact string probably does not exist in Turnitin's database of published content. The AI generated fresh text. It did not copy from an existing source.

This is the core distinction. The Similarity Report catches text that already exists elsewhere. The AI Writing Report catches text that reads like it was machine-generated. Copy-paste is a similarity problem. AI generation is a probability problem.

There is a second layer to how the AI Writing Report works, and it matters for copy-paste scenarios. The model does not analyze everything in your document. The FAQ says:

"The model does not reliably detect AI-generated text in the form of non-prose, or code, nor does it detect short-form/unconventional writing such as bullet points (short non-sentence structures)."

The next sentence is:

"This means that a document containing several different writing types would result in a disparity between the percentage and the highlights."

The FAQ also specifies what does get analyzed:

"This qualifying text includes only prose sentences, meaning that we only analyze blocks of text that are written in standard grammatical sentences and do not include other types of writing such as lists, bullet points (short non-sentence structures), or other non-sentence structures."

The next sentence is:

"This percentage is not necessarily the percentage of the entire submission."

The practical meaning is this. If you copy-pasted a bulleted list, a table of data, or a block of code, the AI model may not even read it. It filters out non-prose content before running its analysis. The Similarity Report, by contrast, reads everything. It will catch a pasted list or a pasted code block just as easily as a pasted paragraph, because it is doing string matching on the whole document.

So when you paste non-prose content from a source, the outcome is asymmetric. The Similarity Report catches it. The AI Writing Report does not, because it never sees it.

Putting the two mechanisms together explains the most common scenario students ask about. You found a paragraph on a website or in a book. You pasted it into your paper. You are worried about both the similarity score and the AI score. Here is what happens.

The Similarity Report will flag the pasted paragraph, assuming the source is in the database. The string matches. The percentage goes up. The source is identified. This is the report doing its job.

The AI Writing Report will not flag the pasted paragraph, assuming the original text was written by a human. The model reads the paragraph as prose, assigns probability scores to its segments, and finds that the word choices look human. The paragraph gets classified as human-written. No AI flag.

This is true even if the pasted text is long. The AI model does not care where the text came from. It cares what the text reads like. A paragraph copied from a 1990s journal article has the word probability patterns of 1990s human academic writing. That is not what AI-generated text looks like.

The FAQ makes the independence explicit. Turnitin says: "The Similarity score and the AI writing detection percentage are completely independent and do not influence each other." A high similarity score from pasted text does not push the AI score up. A high AI score from generated text does not push the similarity score up. They are computed by different systems looking for different things.

If you pasted text from somewhere and you are looking at both reports, here is how to read them.

If the pasted text shows up in the Similarity Report, that is expected. The system found the match. You need to either paraphrase the text in your own words, put it in quotation marks with a citation, or remove it. Quotation marks with a citation is the one case where pasted text is legitimate. Turnitin will still show the match, but your instructor can see it is properly attributed.

If the pasted text is human-written and you are checking the AI Writing Report, you should not see a flag on it. If you do, and the text is genuinely from a human source, the flag is a false positive. The AI model is a probability estimator, not a certainty. It can misclassify. But the more common explanation is that the pasted text is not what got flagged. Read the actual highlighted text. It may be your own writing adjacent to the pasted section, not the pasted section itself.

If you generated text with an AI tool and pasted that in, the situation flips. The Similarity Report probably will not flag it, because the string is new. The AI Writing Report may flag it, because the probability patterns are the ones the model is trained to recognize.

The summary for anyone asking whether Turnitin detects copy and paste:

1. The Similarity Report detects copy-paste. It does string matching against a database. If the source is in the database, the pasted text will show up as a match.
2. The AI Writing Report does not detect copy-paste. It analyzes word probability patterns in prose. Copy-pasted human text reads as human and does not get flagged.
3. The two scores are independent. A high similarity score does not cause a high AI score, and vice versa.
4. If you pasted non-prose content (lists, code, tables), the Similarity Report may catch it but the AI model may not even analyze it, because it only reads prose sentences.
5. If you pasted AI-generated text, the AI Writing Report may flag it even though the Similarity Report will not, because the string is new but the probability patterns are recognizable.

The distinction is not subtle once you know what each system looks for. Similarity checks for existing strings. AI detection checks for probability patterns in prose. Copy-paste is a string problem. AI generation is a pattern problem.