Responsible-use guide
Plagiarism vs text similarity: what's the difference?
A source-matching system can show overlapping wording. Only a contextual human review can decide whether that overlap is properly attributed or amounts to plagiarism.
Similarity, source matching and plagiarism are not synonyms
Text similarity describes wording that overlaps with another text. A source-matching tool compares a submission with material available to its provider and reports candidate sources. Plagiarism is a judgement that someone presented words or ideas without the acknowledgement required in that context.
The first two are observations produced by comparison. The third depends on quotation, citation, disciplinary convention, assignment rules, materiality and sometimes intent. Software does not have all of that context.
| Term | What it describes | Who or what determines it |
|---|---|---|
| Text similarity | Wording that is the same or sufficiently close | A comparison process can measure it |
| Source match | A candidate source associated with overlapping text | A matching system identifies it; a person verifies it |
| Plagiarism | Use of words or ideas without the acknowledgement required by the context | A person applies the relevant academic or editorial standard |
What a similarity-style score does not decide
A percentage summarises a report, but it does not explain why each passage matched. A high number can include legitimate quotation, references, standard wording or a supplied template. A low number can miss unattributed ideas, inaccessible sources or translated and paraphrased material.
Why zero is not proof of originality
A source-matching service can compare only with material it can find and process. Private documents, print-only sources, inaccessible pages, different languages and ideas expressed in new wording may not appear.
Why a high score is not proof of misconduct
The report may include a bibliography, title page, assignment prompt, statutory language, common definitions, correctly marked quotation or the writer's own previously published text. Each passage needs context.
How to review a source match
- 01
Verify the source
Open the reported page and confirm that the matching text appears there. Genutext marks inaccessible candidates as unverified rather than treating them as confirmed matches by default. - 02
Locate the exact overlap
Read complete sentences on both sides. A shared title or technical phrase has a different significance from a copied paragraph. - 03
Check quotation and citation
Look for quotation marks, block formatting, an in-text citation and an accurate reference. Different disciplines have different conventions. - 04
Consider common or supplied language
Identify assignment templates, standard methods, legal wording and unavoidable terminology before treating overlap as authored copying. - 05
Assess materiality and policy
Decide whether the overlap affects the original contribution and whether it breaches the rule that applied to the work.
AI detection asks a different question
| Review dimension | AI detection | Source matching |
|---|---|---|
| Primary question | Does the prose resemble patterns associated with AI-generated text? | Does the wording overlap with a source available to the provider? |
| Primary output | AI-oriented score, label and possible sentence signals | Similarity-style score and source candidates |
| Possible result with novel AI text | May produce an AI signal | May show no source match |
| Possible result with copied human text | May produce a low AI signal | May show a strong source match |
| Human question | What was the writing process and was assistance permitted? | Was the source use acknowledged and acceptable? |
A document can trigger either, both or neither check. Calling generated text “plagiarism” without reference to the applicable rule can blur two distinct concerns: unauthorised assistance and unattributed source use.
A responsible two-check workflow
- Read the task's rules for sources, collaboration, AI assistance and disclosure
- Run the source check only on text suitable for the selected document limit
- Review every meaningful confirmed match and keep unverified candidates separate
- Use AI detection only as a signal about writing patterns, not as a substitute source check
- Verify factual claims and references regardless of both scores
- Seek notes, drafts and a writer explanation if authorship or policy compliance is in question
- Record the source judgement, AI result and final human decision as separate parts of the review
Genutext's combined paid mode puts both reports in one workflow, but the reviewer must preserve the distinction. See the plagiarism checker page for current availability and limits.