{
  "id": "plagiarism",
  "title": "Plagiarism",
  "lang": "en",
  "source_lang": "en",
  "summary": "Plagiarism is the presentation of another person’s words, ideas, or work as one’s own without adequate acknowledgment.",
  "category": "philosophy",
  "subcategory": "ethics",
  "type": "concept",
  "aliases": [
    "Unacknowledged appropriation"
  ],
  "generated_at": "2026-11-02",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/plagiarism/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/plagiarism/",
      "json": "https://aiwiki.page/zh/plagiarism.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Plagiarism is the presentation of another person’s words, ideas, discoveries, creative expression, or other intellectual work as one’s own without adequate acknowledgment. It can involve verbatim copying, close paraphrase, uncredited translation, appropriation of an argument or design, or the submission of work produced by someone else. Its defining concern is misleading attribution: a reader, examiner, employer, or audience is led to misunderstand who contributed the material. Although plagiarism is often discussed alongside intellectual property, it is not identical to infringement of a legal right.\n\nThe concept is important in education, research, journalism, and the creative professions because these activities depend on reliable accounts of authorship and contribution. Acknowledgment enables claims to be checked, allows earlier work to receive credit, and distinguishes an author’s contribution from inherited material. Standards nevertheless vary with genre and setting: a research article, a newspaper report, a traditional song, and an authorized institutional speech do not make the same promises about individual authorship.\n\nPlagiarism consequently has ethical, historical, institutional, legal, and technological dimensions. Its assessment requires more than identifying similar wording. Relevant questions include whether material came from an earlier source, whether attribution was adequate, what expectations governed the work, and whether any overlap resulted from shared conventions rather than appropriation."
    },
    {
      "heading": "Definition, attribution, and ethical significance",
      "text": "Plagiarism is best understood as a failure of attribution within a practice that assigns significance to intellectual contribution. The appropriated material may be a paragraph, an unpublished argument, a diagram, a computer program, a musical passage, or an experimental design. Publication is not a prerequisite: taking an idea from a confidential manuscript or a colleague’s presentation may raise the same attribution problem as copying a printed book. Nor must the source have an identifiable individual author. An organization, research team, or anonymous writer can produce material whose origin should be acknowledged.\n\nAttribution has several components. A citation identifies a source, while quotation marks or block formatting distinguish borrowed wording from the surrounding prose. These functions are related but not interchangeable. A reference at the end of a paragraph may identify the origin of an idea without making clear that several sentences were copied exactly. Conversely, quotation marks around a passage without any usable source information may show that the wording is borrowed while concealing whose wording it is. Adequacy depends on whether readers can understand both the provenance and the extent of the borrowing.\n\nThe principal ethical concerns involve honesty, fairness, and responsibility. In ethics, a deontological account emphasizes duties not to misrepresent authorship; a consequentialist account emphasizes damage to trust, assessment, and knowledge production; and virtue ethics examines habits of intellectual honesty and fairness. These approaches need not yield identical judgments about every case. They nevertheless explain why an accurate statement can be improperly presented: the statement’s truth does not resolve whether its authorship has been represented honestly.\n\nThe issue also belongs to social epistemology, which studies the social conditions of knowledge. Attribution creates a record of intellectual dependence. It allows a reader to distinguish an observation from a secondhand report, trace a disputed claim, and identify who can answer questions about evidence. Unacknowledged borrowing can obscure that record even when no author loses income. It may also redistribute professional rewards, such as grades, appointments, prizes, or research credit, on a false account of contribution.\n\nIntent matters, but institutions do not define its role uniformly. Some reserve the strongest accusation for knowing or reckless appropriation; others classify inadequate attribution as plagiarism even when it results from misunderstanding. An error in note-taking, a mistaken citation, and a deliberate attempt to conceal copying therefore may receive different treatment. A sound distinction separates the observable attribution defect from judgments about responsibility and sanction. Similarity alone establishes neither dishonesty nor an attribution failure, while an attribution failure does not by itself establish deliberate deception."
    },
    {
      "heading": "Historical development of authorship norms",
      "text": "The English term derives ultimately from the Latin plagiarius, associated with kidnapping or the unlawful enslavement of a person. In the first century CE, the Roman poet Martial used the vocabulary of human theft to attack someone who passed off his verses. His epigrams show that complaints about stolen literary credit existed in the Roman Empire, long before modern copyright statutes or university misconduct procedures. They do not, however, demonstrate that ancient authors shared all contemporary assumptions about originality, ownership, or the proper extent of acknowledgment.\n\nAncient and medieval literary cultures combined criticism of deceptive appropriation with extensive practices of imitation and reuse. Rhetorical education encouraged writers to study and emulate exemplary works. Medieval compilers assembled passages from authoritative texts, sometimes identifying sources and sometimes relying on conventions that presumed recognition. Religious commentary could derive authority from continuity with earlier teaching rather than from a claim to unprecedented invention. These practices cannot be classified simply by applying the citation requirements of a modern research journal; their meaning depended on what authors and audiences understood the work to be doing.\n\nDuring the Renaissance, humanist scholarship placed considerable importance on recovering texts, comparing manuscripts, and evaluating the relationship between ancient models and new compositions. Printing enlarged the circulation of texts and made both attribution and unauthorized reproduction more commercially consequential. Printed title pages, publishers’ privileges, and booksellers’ arrangements helped organize the identification and control of works. Yet imitation remained central to literary achievement, and authors could praise creative transformation while condemning a rival’s allegedly servile borrowing.\n\nFrom the eighteenth century onward, influential European accounts of literary originality increasingly emphasized the author’s distinctive creative contribution. The Statute of Anne, enacted in Britain in 1710, was a major development in the history of copyright, but legal protection and ethical judgments about plagiarism followed different trajectories. A work could attract condemnation for unacknowledged borrowing without violating the statute. Conversely, unauthorized reproduction could violate a right even where the copier accurately named the author. The gradual professionalization of scholarship further strengthened expectations that sources, discoveries, and methods should be identified systematically.\n\nModern plagiarism rules developed alongside mass schooling, specialized research, standardized examinations, and professional publishing. Bibliographic conventions made it increasingly feasible to distinguish original contribution from dependence on previous work. These developments were neither simultaneous nor uniform across societies and disciplines. The historical record therefore supports two propositions at once: objections to falsely claiming another’s work are very old, while the particular mechanisms used to define, document, and punish plagiarism are historically variable. Statements that plagiarism was either wholly unknown before modernity or judged identically in every period obscure that distinction."
    },
    {
      "heading": "Forms of appropriation and difficult boundaries",
      "text": "The clearest textual form is verbatim copying without adequate quotation and acknowledgment. It may involve a complete work or only a short but distinctive passage. A related form, often called mosaic plagiarism, combines fragments from several sources with minor substitutions or rearrangements. The resulting text may not reproduce any one source continuously, but it can still conceal substantial dependence. The quantity copied matters to severity, yet a small amount can be significant when it contains a work’s central interpretation, a distinctive formulation, or the answer to an assessed task.\n\nParaphrase does not remove the obligation to identify a source. Replacing words with synonyms while preserving sentence structure and argumentative sequence may reproduce expression too closely; substantially recasting the wording may still appropriate an idea if no acknowledgment is provided. Uncredited translation presents a similar problem. A passage translated from another language can misrepresent authorship even though it contains none of the source’s original words. The relevant relationship is intellectual dependence, not merely matching character strings.\n\nIdeas, data, and nonverbal materials introduce additional forms. A researcher may adopt an unpublished hypothesis from a manuscript received for evaluation; a designer may reproduce a distinctive arrangement; a programmer may submit another person’s code after changing variable names. Appropriation can also concern the architecture of a work: its unusual sequence of examples, original classification, or distinctive synthesis. Such cases require evidence beyond general thematic resemblance. Independent researchers can reach similar conclusions, and commonplace arrangements or standard technical solutions do not automatically establish copying.\n\nContract cheating overlaps with plagiarism but is not coextensive with it. A student who commissions a new essay and submits it as personal work misrepresents authorship even if the writer created every sentence specifically for that commission. No previously published source need exist. By contrast, unauthorized assistance on an examination may violate assessment rules without involving another person’s authored material in the usual sense. Separating these categories helps explain why a plagiarism checker cannot detect every form of academic dishonesty.\n\nReuse of one’s own work is usually called text recycling or, more controversially, self-plagiarism. The phrase differs from ordinary plagiarism because the underlying authorship is not falsely assigned to another person. The concern is instead an inaccurate representation of novelty, separate publication, or independent fulfillment of an assignment. Repeating a standard description of a laboratory method is not necessarily equivalent to republishing the same results as a new study. Similarly, common knowledge ordinarily requires no source citation, but its scope varies by audience: a fact familiar to specialists may require explanation and documentation in a general reference work. No universal word count or percentage resolves these boundaries."
    },
    {
      "heading": "Relationship to copyright and other legal rights",
      "text": "Plagiarism and copyright infringement answer different questions. Plagiarism concerns acknowledgment and representations of intellectual contribution; copyright regulates specified uses of protected expression under the applicable legal system. An author can plagiarize material whose copyright has expired, because freedom to reproduce it does not authorize a false claim of authorship. Conversely, a person can accurately identify a living author and still reproduce protected material without permission in circumstances that constitute infringement. Attribution is therefore neither a universal substitute for permission nor a guarantee of legal compliance.\n\nCopyright generally distinguishes an idea from its particular expression. Scientific principles, historical facts, and abstract methods are not protected in the same manner as an author’s wording or an original illustration. Academic attribution norms can extend further: presenting another researcher’s distinctive explanation or hypothesis as a personal discovery may constitute plagiarism even when the underlying idea is not copyrightable. This difference is especially important in research, where credit for conceptual advances may carry greater significance than ownership of a particular paragraph.\n\nMaterial in the public domain provides a straightforward example. An out-of-copyright poem may ordinarily be reproduced without obtaining copyright permission, but submitting it as a newly written personal composition remains an attribution violation. Open licensing creates another distinction. A license may permit redistribution or adaptation subject to conditions, including credit. Failure to meet those conditions can raise legal issues as well as concerns about attribution. Compliance with a license, however, does not necessarily satisfy the separate authorship expectations of a school assignment or research publication.\n\nSome legal systems recognize moral rights, including rights connected with attribution and the integrity of a work. Article 6bis of the Berne Convention addresses an author’s right to claim authorship and to object to certain prejudicial modifications. The scope, duration, transferability, and enforceability of these rights vary by jurisdiction. In the United States, fair use can permit some uses of copyrighted material without permission, but its application involves a contextual legal analysis rather than a fixed rule about a permissible number of words. Other jurisdictions provide differently structured exceptions.\n\nThere is no single worldwide offense called plagiarism with a uniform legal definition. A dispute described publicly as plagiarism may instead be litigated through copyright, contract, fraud, employment, or other legal doctrines, depending on the facts and jurisdiction. Institutional sanctions can also operate independently of a court judgment: a journal may reject a manuscript or a university may discipline a student even when no copyright claim exists. The fact that an act is lawful does not settle whether it meets professional standards, just as an ethical accusation does not by itself establish legal liability."
    },
    {
      "heading": "Education, assessment, and institutional procedure",
      "text": "In higher education, plagiarism rules protect the relationship between submitted work and the learning or competence that it is supposed to demonstrate. An essay may assess interpretation, a laboratory report the analysis of observations, and a programming exercise the ability to construct an algorithm. If the central intellectual work was performed by someone else and this dependence is concealed, the resulting grade does not measure the intended achievement. The problem is therefore not limited to depriving a source author of credit; it also concerns the validity and fairness of assessment.\n\nThe boundaries depend partly on the assignment. Collaboration may be required in one course and prohibited in another. A group project can legitimately have shared authorship, whereas an individual examination may permit no consultation. Editing, translation assistance, tutoring, and software support likewise have different implications according to their permitted scope. Proofreading that corrects punctuation is not equivalent to rewriting an argument. A bibliography cannot cure a breach of a requirement that the student perform a particular task independently, because source acknowledgment and authorization to receive assistance are separate matters.\n\nResearch on student writing has examined patchwriting, a practice in which a writer stays close to a source’s language while substituting words or making limited grammatical changes. Rebecca Moore Howard’s 1993 article on the subject helped establish its importance in debates about novice academic writing. Patchwriting can reflect an incomplete command of disciplinary language or source integration rather than a calculated scheme to deceive. It can nevertheless result in unacceptable borrowing in a submitted text. Its educational interpretation therefore requires attention to the writer’s development, the assignment, and the institution’s published rules.\n\nAcademic integrity procedures commonly distinguish an initial concern from an established finding. Relevant evidence can include the submitted work, likely sources, instructions, drafts, notes, and explanations of the writing process. Fair procedures make the allegation intelligible, allow a response, and apply the institution’s stated evidentiary standard. A similarity report is one item of evidence rather than a substitute for this process. Differences in the amount appropriated, the significance of the material, prior instruction, and evidence of concealment may affect how a case is classified.\n\nInstitutional responses range from instructional correction and resubmission to grade penalties, suspension, or, in serious established cases, revocation of a qualification. Not every institution offers all these responses, and local regulations govern their use. Prevention also operates through curriculum and assessment design: explicit instruction in quotation and paraphrase, clear collaboration rules, staged submissions, and tasks connected to demonstrable research processes. These measures address different causes of misconduct or error. They also clarify that acceptable academic writing involves accountable use of sources, not simply producing a document with a low similarity score."
    },
    {
      "heading": "Research credit and scholarly publishing",
      "text": "In research, attribution helps allocate credit while preserving a verifiable account of how results developed. A publication distinguishes earlier findings from new observations, methods, interpretations, and analyses. Because funding, employment, and scientific priority can depend on that distinction, plagiarism may distort both the scholarly record and professional reward. The United States federal framework identifies plagiarism, together with fabrication and falsification, as a category of research misconduct. Its definition includes appropriation of another person’s ideas, processes, results, or words without appropriate credit; findings under that framework also depend on specified standards concerning conduct and evidence.\n\nResearch plagiarism may occur before publication. A reviewer can encounter an unpublished method through peer review, or a grant evaluator may learn of a proposed study in a confidential application. Using such information without authorization can violate confidentiality as well as attribution expectations. These cases may be difficult to establish because the materials were not publicly available and intellectual development can proceed along parallel paths. Dated drafts, correspondence, submission records, and laboratory documentation can therefore be important in distinguishing independent work from appropriation.\n\nPlagiarism is related to, but different from, authorship disputes. Excluding a person who made a qualifying contribution, adding an honorary author, or using an undisclosed writer may breach authorship standards without fitting every definition of plagiarism. Citation also cannot replace authorship where a person’s contribution warrants a place on the author list. Conversely, listing someone as an author does not identify which passages or findings originated in unrelated prior publications. Modern contributor statements can clarify roles such as conceptualization, data collection, analysis, and writing, although role labels do not eliminate disagreements about credit.\n\nDuplicate publication and undisclosed text recycling can misrepresent the novelty or independence of the literature. If substantially the same dataset appears in multiple articles without transparent cross-reference, a later evidence synthesis may mistakenly treat the reports as separate studies. In clinical research, such duplication can affect the apparent weight of evidence concerning an intervention. Transparent secondary publication, translated publication, or follow-up analysis may be legitimate when the relationship to the original work is disclosed and relevant editorial requirements are met. The decisive issue is not repetition alone but what the new publication claims to contribute.\n\nEditorial responses depend on the extent and consequences of the problem. A manuscript can be rejected before publication; a published article may require a correction or retraction. Retraction notices identify why a work should no longer be relied upon in its published form, although retraction is not always a finding of author misconduct. Plagiarism also intersects with reproducibility: concealed dependence can obscure the actual source of a method or dataset and frustrate attempts to reconstruct the work. Reliable publishing practice therefore concerns provenance as well as the surface accuracy of statements."
    },
    {
      "heading": "Literature, journalism, and creative practice",
      "text": "In literature, borrowing ranges from concealed appropriation to openly signaled engagement with earlier works. Allusion, parody, adaptation, quotation, and pastiche can make the relationship to a predecessor part of the work’s meaning. Intertextuality describes relationships among texts more broadly; it does not by itself excuse every unacknowledged borrowing. The relevant question is often what a reasonably informed audience is invited to believe. A recognizable quotation placed in a new context can function differently from an obscure passage reproduced as evidence of the new author’s original verbal achievement.\n\nGenre conventions affect that interpretation. A historical novel may incorporate public documents, while an annotated scholarly edition is expected to identify textual sources explicitly. Traditional tales and communal repertoires complicate assumptions that every work has a single originating author. In oral tradition, variation and transmission are often constitutive features rather than departures from a fixed original. Even so, practices of community acknowledgment, custodianship, or attribution may matter. Absence of modern individual ownership does not mean that all uses are regarded as socially equivalent.\n\nJournalism distinguishes the use of information from the appropriation of reporting and prose. Reporters can independently describe the same event, and routine factual statements may resemble one another without copying. However, reproducing another outlet’s distinctive reporting, interview material, or narrative passages without suitable acknowledgment can mislead readers about how the report was produced. Material supplied by news agencies, press releases, or authorized syndication arrangements introduces additional conventions. Attribution and disclosure depend on the relationship, editorial policy, and whether the resulting presentation implies independent reporting that did not occur.\n\nIn music, visual art, film, and architecture, disputes may concern melody, composition, images, sequences, or designs rather than prose. Influence alone is not plagiarism, and shared styles or standard forms are not equivalent to copying an identifiable work. A recurring chord progression, a conventional portrait pose, or a standard building component may be too general to establish meaningful appropriation. A distinctive combination or passage can raise a different question. Legal determinations about protected expression and substantial similarity must still be distinguished from critical judgments about originality and acknowledgment.\n\nGhostwriting further shows why undisclosed individual authorship is not always treated uniformly. Speeches, executive statements, celebrity memoirs, and institutional documents may be produced through authorized collaboration. An audience may understand a speech delivered by an officeholder as an authorized statement rather than a claim that every sentence was personally composed. A student essay ordinarily carries a different representation. These differences do not make authorship irrelevant; they show that plagiarism is assessed against the purposes and expectations of the practice, including what a signature, byline, or submission declaration communicates."
    },
    {
      "heading": "Detection, textual evidence, and its limitations",
      "text": "Plagiarism detection begins with identifying a potentially meaningful relationship between works. Traditional investigation used close reading, bibliographic knowledge, and comparison with likely sources. Abrupt changes in terminology, style, citation format, or level of technical expertise could prompt further inquiry, but none is conclusive. A writer may change style when summarizing a specialist source, and an unusually polished passage may simply have received extensive revision. Evidence becomes stronger when a suspected source contains the same distinctive wording, unusual sequence, or inherited error.\n\nComputer-based systems automate parts of this comparison. A common method divides documents into short sequences of words or characters, often called n-grams, and compares selected representations against indexed material. Other algorithms examine longer alignments, edit distance, document structure, or semantic similarity. Systems can search web pages, publications, and repositories of prior submissions, subject to their access and contractual arrangements. Their output usually identifies overlapping passages and possible source documents rather than establishing the circumstances under which the overlap arose.\n\nA similarity percentage is a measurement produced by particular settings and a particular comparison collection. It is not the probability that a writer plagiarized. Quoted passages, references, assignment questions, legal formulas, and standard methods can increase overlap without misconduct. Excluding bibliographies or short matches can substantially alter a score. Conversely, an essay that appropriates an unpublished idea or uses an unindexed source may receive a low score. There is therefore no context-free numerical threshold that reliably separates legitimate writing from plagiarism.\n\nDetection also encounters the problem of direction and dependence. Two documents may share wording because one copied the other, both copied a third source, both used an authorized template, or both describe a standard procedure. Publication dates help but may not reveal when a draft was written or circulated. Stronger investigations examine distinctive correspondences, access to the alleged source, chronology, acknowledgment, and alternative explanations. Repeated unusual errors can be particularly informative because they are less likely than ordinary phrases to arise independently, although even this evidence requires contextual interpretation.\n\nCross-language borrowing and substantial paraphrase are harder to identify than direct copying. Machine-learning methods can compare representations of meaning, but semantic resemblance alone is weak evidence of appropriation: independent discussions of the same topic necessarily overlap in concepts. Detection services also raise data privacy questions when student papers, unpublished research, or confidential documents enter external databases. Retention rules, access controls, and the terms under which documents are reused affect the acceptability of such systems. Detection is thus both an evidentiary activity and an institutional arrangement for handling intellectual work."
    },
    {
      "heading": "Generative artificial intelligence and digital authorship",
      "text": "Digital networks have made copying, searching, and comparing texts easier, while artificial intelligence has complicated the relationship between textual originality and authorship. A large language model generates text using patterns learned from training data. Its output may contain newly assembled wording, common expressions, or, in some circumstances, memorized passages. These possibilities raise different questions. A generated passage need not reproduce an identifiable source to be improperly submitted as an unaided personal performance, and generated wording that does reproduce a source can still require attribution.\n\nUndisclosed use of a generative system is therefore not automatically synonymous with plagiarism. In an assessment that prohibits such assistance, it may instead be classified as unauthorized assistance, misrepresentation, or another breach of academic rules. In a setting that permits and discloses its use, the same technical process may be acceptable. The distinction depends on what the task requires the named author to do and what the final document claims about its production. Acknowledging software use does not, by itself, establish that every underlying source has been properly credited.\n\nThe relationship between training and output must also be kept separate. Whether particular uses of protected works in model development comply with copyright is a legal question distinct from whether a particular generated response plagiarizes a source. A model’s statistical dependence on a large collection does not mean that every output is a quotation from that collection. Conversely, the absence of an easily identifiable source does not establish that a submitted work demonstrates its presenter’s knowledge, reasoning, or writing ability. Legal provenance, textual dependence, and assessment validity are different dimensions.\n\nGenerated references create a further problem for attribution. Language models can produce plausible bibliographic details that do not correspond to real publications or can attach a real source to a claim it does not support. Such references fail the evidentiary purpose of citation even if the surrounding prose is not copied. In research and professional publication, responsibility for the accuracy of claims and references remains with the accountable human authors. A system cannot take responsibility for an investigation, answer an allegation, or meet the obligations normally attached to scholarly authorship.\n\nTools marketed as detectors of AI-generated text generally infer patterns associated with machine output rather than establishing copying from a source. Their results are consequently different from source-matching evidence and can include both false positives and false negatives. A finding that prose resembles model output does not establish plagiarism, identify an uncredited author, or show whether a permitted tool was used. Policies addressing digital authorship therefore distinguish source attribution, disclosure of assistance, confidentiality, originality requirements, and responsibility for the final work rather than treating them as a single question."
    }
  ],
  "links": [
    "intellectual-property",
    "education",
    "ethics",
    "deontology",
    "utilitarianism",
    "virtue-ethics",
    "knowledge",
    "latin",
    "roman-empire",
    "renaissance",
    "printing",
    "translation",
    "language",
    "copyright",
    "public-domain",
    "moral-rights",
    "united-states",
    "fair-use",
    "higher-education",
    "research-misconduct",
    "peer-review",
    "reproducibility",
    "literature",
    "intertextuality",
    "oral-tradition",
    "music",
    "art",
    "film",
    "architecture",
    "algorithm",
    "machine-learning",
    "data-privacy",
    "artificial-intelligence",
    "large-language-model",
    "training-data"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}