plagiarism
Enable an AI agent to record and assess suspected plagiarism by connecting reused contributions, their presentation, applicable attribution expectations, supporting evidence and proportionate next actions.
Research draft, second pass
A second pass drafted this model: the structure a model of this thing needs, and what is known about it in the world. The line under this one says how the second half was obtained - researched against sources, or recalled without web access, in which case nothing here was read anywhere and every claim is a lead to verify. Unreviewed either way.
Researched by: Codex + Grok
Purpose and description
Enable an AI agent to record and assess suspected plagiarism by connecting reused contributions, their presentation, applicable attribution expectations, supporting evidence and proportionate next actions.
Plagiarism is the unattributed appropriation of another identifiable person's words, ideas, processes, results, or work products in a setting where original authorship is legitimately expected, done so as to take credit or other benefit that is not given to the source.
It can be Locate and compare the precise contributions alleged to have been reused.; Trace candidate sources and test shared-source or independent-creation explanations.; Assess acknowledgment against a recorded contextual requirement.; Prepare an evidence-linked account that separates observations, allegations and judgments.; Request missing source material, explanations or authorized review.; Propose and track attribution corrections, disclosure or other remedies within the agent's authority..
Distinguishing features
Textual or structural similarity alone does not establish plagiarism: the model must connect an identifiable contribution to its source, its presentation and an applicable attribution expectation.
Properly acknowledged reuse may resemble plagiarism in a similarity report while failing the test of missing or misleading acknowledgment.
Permission to reuse and acknowledgment of origin are separate tests; satisfying one does not establish that the other is satisfied.
Independent creation, common knowledge and material inherited from a shared source must remain alternatives until evidence distinguishes them from borrowing.
Reuse of one's own work must be assessed against a specific disclosure or originality requirement rather than automatically treated as appropriation of another person's work.
Scope
+ The submitted, published or otherwise presented work and the specific passages, ideas, methods, code, images or other contributions under examination.
+ Candidate source contributions and evidence of borrowing, derivation or shared origin.
+ Attribution, quotation, acknowledgment and prior-publication disclosures as encountered by the intended audience.
+ The contextual rules and exceptions used to assess whether the presentation constitutes plagiarism.
+ The allegation, assessment, response and correction history of the specific act.
- Copyright ownership, licensing and infringement determinations.
- General authorship allocation and contributor-credit disputes without an alleged misrepresentation of borrowed contributions.
- Fabrication, falsification and broader research-integrity investigations.
- Contract cheating, impersonation and unauthorized assistance as independent misconduct categories.
- Operation and benchmarking of similarity-detection systems.
- Institution-wide disciplinary procedures and sanctions beyond their application to this case.
Characteristics
- Contribution under examination
- Wording; idea or argument; method; data selection or arrangement; code; image or design; mixed; unresolved Determines what must be compared and what acknowledgment could meaningfully identify the borrowed contribution.
- Source relationship
- Links a bounded contribution in the presented work to one or more candidate antecedents, intermediaries or common sources. Prevents an earlier-looking match from being treated automatically as the actual source.
- Transformation of borrowed material
- Verbatim reuse; close paraphrase; translation; rearrangement or mosaic assembly; adaptation across media; conceptual reuse; unresolved Makes nonverbatim forms of possible appropriation visible.
- Acknowledgment condition
- Absent; present but ambiguous; source identified but reused extent unclear; apparently sufficient under the identified rule; disputed; unassessed Records whether readers can distinguish the presenter's contribution from material originating elsewhere.
- Extent of implicated material
- Count or proportion with explicit unit, denominator and comparison method, such as words, passages, functions or figures; unknown when not defensibly measurable Supports an assessment of scope without equating a similarity percentage with a plagiarism determination.
- Importance to the claimed contribution
- Incidental; supporting; central to the claimed contribution; disputed; unassessed, with a case-specific justification A small borrowed component may carry the work's main claim or claimed originality.
- Applicable attribution expectation
- Links the act to an identified assignment instruction, publication policy, professional convention or other governing expectation and its applicable version. Makes the basis for judgment explicit and open to challenge.
- Assessment status
- Unassessed; suspected; under review; supported under an identified rule; not supported; inconclusive; contested Separates detection signals, allegations and determinations.
Also called
Where this came from
wikidata · CC0 1.0
Also registered as vr.tr.plagiarism
Drafted structure
Bundle to layer to finding to question, as the second pass will find it: 6 bundles · 11 layers · 18 findings · 33 questions.
Borrowed contribution Identifies what is alleged to have been taken and where it appears in the presented work.
A plagiarism assessment needs a bounded contribution, not merely a suspicious document or aggregate match score.
Implicated material
Locates the material and identifies the kind of contribution at issue.
Contribution boundary
Record the specific expression, idea, method or artifact component whose origin or acknowledgment is disputed.
- Which exact passages, claims, code units, images or other components are implicated, and in which version of the work? boundary
- Is the allegation about copied expression, an appropriated idea, a method or another identifiable contribution? definition
Reuse form and weight
Characterizes transformation, extent and importance without treating them as automatic verdicts.
Reuse profile
Record how the material was transformed and how much of the work's claimed contribution depends on it.
- Does the comparison indicate verbatim copying, paraphrase, translation, mosaic assembly, adaptation or conceptual reuse? definition
- What is the extent of the implicated material, using an explicit unit and denominator where appropriate? measurement
- How does the implicated material support the work's central claim or claimed originality? boundary
Source and derivation Examines where the contribution originated and whether the presented work plausibly derives from it.
A matching work may be an intermediary, a later copy or another descendant of a common source.
Source lineage
Distinguishes candidate originators, intermediaries and available versions.
Antecedent evidence
Record inspected source material, chronology and limits on what can be inferred about origin.
- Which candidate sources have been inspected, and which exact versions contain the relevant contribution? provenance
- What evidence establishes their chronology and distinguishes an originating source from an intermediary or later reproduction? provenance
Derivation alternatives
Tests borrowing against explanations that can produce similarity without the alleged appropriation.
Borrowing versus coincidence
Record evidence for derivation alongside plausible shared-source, conventional-form and independent-creation explanations.
- Which distinctive correspondences support derivation beyond generic wording or routine structure? measurement
- Could a shared source, required template, standard expression or independent development explain those correspondences? boundary
- What evidence of access or transmission exists, and what remains unknown? provenance
Presentation and attribution Examines what the audience was led to understand about the contribution's origin.
The source relationship must be assessed together with the acknowledgment actually available to readers.
Audience-facing credit
Records citations, quotation boundaries and other signals of borrowed material.
Acknowledgment coverage
Record whether acknowledgment identifies both the source and the scope of the reused contribution.
- What citation, quotation, caption, code notice or acknowledgment accompanies the implicated material in the examined version? provenance
- Can the intended audience identify where the borrowed contribution begins and ends, including any verbatim wording? boundary
Originality and disclosure
Records explicit or contextual claims of originality and relevant prior-use disclosures.
Claimed origin
Identify what the presenter claimed as their own or as new, including cases involving their prior work.
- What explicit statement or submission context presents the implicated contribution as the presenter's own or newly produced work? provenance
- If the contribution comes from the presenter's prior work, what prior-use disclosure was made and what specific disclosure requirement applies? boundary
Contextual assessment Connects the evidence to the attribution expectations and decision authority governing the act.
The model must expose why an act meets or fails a particular plagiarism standard instead of assuming a universal threshold.
Governing expectations
Identifies applicable requirements and their boundaries.
Attribution rule
Record the rule used to evaluate the act, including relevant exceptions and treatment of intent.
- Which identifiable policy, instruction or convention governed attribution when and where the work was presented? provenance
- How does that standard treat common knowledge, routine expressions, authorized collaboration and reuse of prior work? boundary
- Does the standard make intent relevant to classification, consequences or both, and what evidence addresses it? definition
Evidence to determination
Separates comparison results, allegations, responses and authorized conclusions.
Supported assessment
Record which elements of the applicable standard are supported, disputed or unresolved.
- Which required elements are supported by inspected evidence, and which rely only on a detector output or an unverified allegation? measurement
- What explanation or counterevidence has the person concerned supplied, or is their response still unavailable? provenance
- Who may make the determination, and what assessment status is justified pending that decision? action
Correction and case resolution Connects the identified attribution problem to possible remedies and records their effects.
Recognizing plagiarism must support bounded action while preserving the distinction between correcting a work and resolving a misconduct finding.
Remedy selection
Identifies actions that address the specific borrowed contribution and misleading presentation.
Attribution repair
Record proposed remedies, their justification and the authority needed to carry them out.
- Would corrected citation, quotation boundaries, contribution credit or prior-use disclosure address the identified presentation problem? action
- What further review or action is required if attribution repair cannot resolve the issue, and who is authorized to undertake it? action
Resolution verification
Tracks corrected versions and unresolved consequences of the original presentation.
Correction versus closure
Record what changed in the work and whether the allegation or determination remains open.
- Which implicated components were corrected in which version, and what inspection confirms that the correction addresses them? provenance
- Does the authorized reviewer regard the case as resolved, contested or still open despite the correction? action
- Which circulated versions or affected audiences still require an authorized correction or notice? boundary
Evidence and external alignment What the world already says about this thing, gathered so the model can be checked against it.
A model that cannot be lined up against existing standards, identifiers and practice cannot be adopted by anyone who already uses them.
Reported evidence
Findings from the breadth pass, kept separate from the structural claims.
Kinds and varieties
Reported by the breadth pass; each item needs checking against its source before it becomes normative.
- Verbatim (direct/clone) plagiarism: word-for-word copying without quotation and attribution
- Mosaic plagiarism / patchwriting: stitching phrases or sentence structures from one or more sources into ostensibly original text
- Paraphrase plagiarism: restating another's ideas while keeping their content, and often their structure, without credit
- Idea and structure plagiarism: taking concepts, arguments, or organisation without taking the wording
- Translation plagiarism: rendering a source from another language as original work
- Self-plagiarism / text recycling: reusing one's own previously submitted or published material as if it were new
- Contract cheating / ghostwriting (global or complete plagiarism): submitting an entire work produced by someone else, paid or unpaid
- Source-based plagiarism: invented, unread, secondary-as-primary, or otherwise misleading citations
- Which of these kinds and varieties hold for the sense of plagiarism this model covers, and on what evidence? provenance
Identifiers and schemes
Reported by the breadth pass; each item needs checking against its source before it becomes normative.
- ISO 5127 vocabulary term - plagiarism (as defined in ISO 5127 / ČSN ISO 5127-2003: presenting another author's intellectual work, borrowed or imitated in whole or in part, as one's own) - A documentation-vocabulary entry, not a taxonomic code; the 2003 Czech adoption is the edition cited in the source actually read. Wikidata Q-id was not confirmed in this pass and is omitted.
- US Public Health Service research-misconduct category - plagiarism within FFP (fabrication, falsification, plagiarism), 42 CFR Part 93 as administered by the Office of Research Integrity - ORI's operational wording: appropriation of another person's ideas, processes, results, or words without giving appropriate credit.
- Which of these identifiers and schemes hold for the sense of plagiarism this model covers, and on what evidence? provenance
Standards and regulation
Reported by the breadth pass; each item needs checking against its source before it becomes normative.
- US Department of Health and Human Services, Office of Research Integrity - 42 CFR Part 93 (Public Health Service Policies on Research Misconduct), treating plagiarism as one of the three core research-misconduct offences
- Committee on Publication Ethics (COPE) - position statements and flowcharts on suspected plagiarism in submitted and published articles, and on acceptable levels of duplication
- ISO 5127 (Information and documentation - vocabulary), as nationally adopted e.g. ČSN ISO 5127-2003, which defines the term for documentation practice
- Journal and publisher research-integrity policies that implement COPE, e.g. PLOS Ethical Publishing Practice (screening with Crossref Similarity Check / iThenticate; text-recycling rules)
- Institutional academic-integrity / honour codes and student-misconduct procedures (Stanford, Yale, Princeton and others, as summarised in the Wikipedia survey of campus definitions)
- Copyright, moral-rights, unfair-competition, fraud, and contract law, which can attach to the same facts but do not generally name 'plagiarism' as a statutory offence
- Which of these standards and regulation hold for the sense of plagiarism this model covers, and on what evidence? provenance
Real-world use
Reported by the breadth pass; each item needs checking against its source before it becomes normative.
- Universities screen student coursework with text-matching services (Turnitin and similar) and then adjudicate matches through academic-misconduct panels, with sanctions from a failed assignment to expulsion
- Journals and publishers run submitted manuscripts through Crossref Similarity Check / iThenticate; confirmed plagiarism can mean rejection, correction, or retraction
- Newsrooms treat unattributed lifting as a journalistic-ethics breach; reporters are typically suspended pending investigation and may be dismissed
- Wikipedia moderation (e.g. CopyPatrol plus iThenticate) flags unattributed text and copyright problems on edits
- Research-integrity offices investigate alleged plagiarism of ideas, methods, or wording in grants, theses, and papers
- Students and researchers encounter it as a citation-and-paraphrase skill they are taught, and as a detection report they must interpret
- Which of these real-world use hold for the sense of plagiarism this model covers, and on what evidence? provenance
Typical measurements
Reported by the breadth pass; each item needs checking against its source before it becomes normative.
- Text-similarity / overlap index from matching software - No accepted numeric cutoff; COPE advises against using a percentage as the investigation trigger. Nature, location, and concentration of the match matter more than the score - percent of submitted text matching a reference corpus
- Contiguous duplicated span - From isolated boilerplate phrases (often tolerated in methods) to whole paragraphs or sections (usually investigated) - words or sentences of continuous overlap
- Number of matched source documents - One source (clone/copy) versus many (mosaic/mashup); count alone does not decide the case - count of sources
- Which of these typical measurements hold for the sense of plagiarism this model covers, and on what evidence? provenance
Failure modes and hazards
Reported by the breadth pass; each item needs checking against its source before it becomes normative.
- A similarity percentage is treated as a finding of plagiarism, producing false accusations on methods boilerplate, common phrases, or legitimate thesis-to-paper reuse
- False negatives: paraphrase, synonym substitution, translation, idea theft, contract cheating, and much AI-generated text evade source-matching
- AI-writing detectors used as dispositive proof, despite being non-specific; contested in student-conduct hearings (e.g. Newby v. Adelphi University as reported)
- Detection corpora are stronger for high-resource languages, so work in smaller languages is both less well screened and more easily missed
- Confirmed plagiarism can retract papers, end academic appointments, fail or expel students, and destroy journalistic credibility
- If plagiarism becomes routine in assessed work, the associated qualification no longer certifies the learning it claims to
- Which of these failure modes and hazards hold for the sense of plagiarism this model covers, and on what evidence? provenance
Regional variation
Reported by the breadth pass; each item needs checking against its source before it becomes normative.
- In US and much Anglophone practice, plagiarism is primarily an institutional ethics offence, not a named crime; the same facts may still support copyright, fraud, or contract claims
- Continental European droit d'auteur / moral-rights systems can give the original author a legal remedy for unattributed use even when a US-style copyright analysis would not
- Czech copyright law (as reported) does not define plagiarism, whereas neighbouring systems such as German law are described as using the concept more directly
- Anglo-American assessment culture expects marked originality and citation; some pedagogical traditions treat close imitation as legitimate apprenticeship, which collides with imported integrity codes
- Text-matching tools and publisher workflows are built around large English and other high-resource corpora, so detection practice is uneven by language
- Which of these regional variation hold for the sense of plagiarism this model covers, and on what evidence? provenance
Neighbouring kinds and how to tell them apart
Reported by the breadth pass; each item needs checking against its source before it becomes normative.
- Copyright infringement - Ask whether exclusive rights of a copyright holder were used without licence. Plagiarism is a false claim of authorship or unearned credit; it can occur in public-domain or licensed text, and infringement can occur without any authorship claim.
- Piracy / unauthorised copying - Piracy copies or distributes a work without permission but typically still presents it as the original author's. Plagiarism requires appropriation of authorship.
- Proper quotation and citation - If the borrowed words are marked as quotation (or clearly paraphrased) and the source is given, the originality-and-credit test for plagiarism is not met.
- Common knowledge - Material a typical member of the intended audience would already know without a source need not be cited; the test is audience knowledge, not the writer's memory.
- Cryptomnesia - The writer believes the material is original after unconscious recall. Academic policy may still treat the product as plagiarism; the psychological test is belief, the integrity test is missing credit.
- Duplicate publication / text recycling of one's own work - The source is the same author. It is dishonest about novelty, not theft of another's work; many specialists prefer 'redundant publication' or 'text recycling' to 'self-plagiarism'.
- Authorised ghostwriting - A speechwriter or named contributor working under an agreed disclosure or disciplinary convention is not presenting the work as the ghost's original assessed authorship; a student essay mill is.
- Fabrication or falsification - Those invent or alter data or results. Plagiarism takes existing work; the three are the FFP triad and can co-occur (e.g. fake citations).
- Which of these neighbouring kinds and how to tell them apart hold for the sense of plagiarism this model covers, and on what evidence? provenance
Sources
- Plagiarism. Wikipedia, Wikimedia Foundation - Core definition; distinction from copyright infringement; academic, journalistic, and artistic practice; self-plagiarism and reverse plagiarism; detection and AI; Fishman's five-element test as cited there.
- When I use a word . . . Academic fraud - plagiarism. The BMJ (Jeffrey K. Aronson) - US Office of Research Integrity definition; difficulty of drawing the occurrence line; adapted typology including concealed paraphrasing, purloining, ghostwriting.
- Determining acceptable levels of plagiarism/duplication. COPE Council, Committee on Publication Ethics - No fixed percentage of overlap as an investigation trigger; nature, concentration, article type, and disciplinary norms decide the case.
- Ethical Publishing Practice (Plagiarism). PLOS journals - Publisher operational definition covering text, ideas, images, data, and text recycling; methods-section exception; screening via Crossref Similarity Check / iThenticate; COPE follow-up.
- The 5 Types of Plagiarism. Scribbr - Practice typology used in student-integrity teaching: global, verbatim, paraphrasing, patchwork, and self-plagiarism.
- Taxonomy of academic plagiarism methods. T. Vrbanec and A. Meštrović, Zbornik Veleučilišta u Rijeci / arXiv 2105.12068 - Survey of definitions and a technical taxonomy (clone, translation, copy, substitute, remix, hybrid, mashup, waste, ghost-product).
- Plagiát. Czech Wikipedia, Wikimedia Foundation - ISO 5127 (ČSN ISO 5127-2003) vocabulary definition; note that Czech copyright law does not itself define plagiarism, unlike some neighbouring systems.
- Плагиат. Russian Wikipedia, Wikimedia Foundation - Authorship-appropriation as the obligatory mark of plagiarism, contrasted with copyright piracy that copies without claiming to be the author.
What the second pass must settle
- Which researched definitions should govern the boundary between plagiarism, inadequate citation and disputed authorship across the intended contexts?
- Should undisclosed reuse of one's own work be classified within plagiarism or linked to a neighbouring redundant-publication or disclosure model?
- How should attribution expectations be established for ideas, methods, code, images and oral contributions when no explicit local policy exists?
- What evidence is sufficient to distinguish derivation from independent creation or common-source reuse, especially across languages and media?
- How should AI-mediated reuse and fabricated or misleading citations be represented when the underlying source relationship cannot yet be established?