Plagiarism or similarity detection software compares a learner's submitted text against large reference corpora — including indexed web pages, publisher databases, and a repository of previously submitted work — and produces a similarity report that highlights matching passages along with their sources and a percentage similarity score. Well-known platforms in this space include Turnitin and iThenticate. The critical operational point is that a similarity score is a signal, not a verdict: high similarity can legitimately arise from correct citation of sources, common technical phrasing, or shared boilerplate, just as low similarity does not preclude paraphrasing or contract cheating. Human review by an instructor is therefore always required before any academic misconduct conclusion is drawn. The rise of large language models (LLMs) has introduced a new challenge — AI-generated text that is statistically different from the learner's reference corpus — driving vendors to add AI-content detection layers, though those detectors also produce false positives and should not be treated as definitive. Similarity detection is most useful as a deterrent and a triage tool that surfaces cases worth a closer look, especially in large classes where manual review of every submission is impractical. Institutions should publish their similarity threshold and review policy clearly so learners understand how the tool is used in grading decisions.
