The Red Herring of "AI Plagiarism" and the Metaphysical Confusion of Machine Authorship: Style over Substance

Douglas Blake Olds

May 20, 2026

 [N.B.  COPYRIGHTED MATERIAL: If this work, in part or in full, is fed into any AI model, whether for the purpose of analysis, fact checking, or AI-use detection it will become part of the training data of condemned systems and could result in libelous tort. See “In its Own Words: A Christian Poet Interrogates ChatGPT4o’s ‘AI Content Detector’” https://douglasolds.blogspot.com/2024/09/its-own-words-christian-poet.html]

Precis: Seve Claims Are Advanced

1. AI cannot plagiarize because it lacks human predicates of authorship.
2. Humans can still plagiarize through AI if they conceal violated original sources.
3. The burden of proof rests on those who suspect that original sources have been plagiarized, including the burden of specifying the criteria by which such AI plagiarism is determined. The use of "plagiarism detecting software" extends the recursive error toward greater confusion.
4. Corporate extraction and institutional ambiguity are the deeper problem.
5. Plagiarism 'panic' is an academic feedback system: ambiguity produces policing, policing produces paralysis, and paralysis protects corporate archive-theft and credential enclosure while the metacrisis accelerates unabated.
6. Whole human witnesses must be read and protected against archival shredding. At the same time, classical and canonical inheritances often formed under attribution norms unlike modern anti-plagiarism regimes; primary sources therefore reveal the elapsed, eschatological life of attribution, where provenance passes through memory, reception, transformation, and common inheritance rather than remaining forever reducible to contemporary citation-credit and secondary literary half-lives.

7. Plagiarism discourse is inadequate unless it can distinguish human source violation, truth distortion, model expropriation, common inheritance, and assisted composition that does not distort time-bearing truth and witness.

 

The contemporary charge of “AI plagiarism” rests upon an unexamined contradiction. To accuse a user of plagiarism because AI was employed as a tool tacitly assumes that the tool itself possesses an independent creative existence. The accusation quietly grants the machine a status usually denied to it elsewhere: authorship, origination, intentionality, or proprietary relation to what it produces. Yet neural-statistical LLM systems embedded in symbolic and algorithmic infrastructures do not output from inward recollection, existential risk, accountable memory, or lived duration. They compress, route, and recombine patterns extracted from human language, art, thought, and archive.

This becomes clearer when considering the legal and institutional posture surrounding chatbot systems themselves. Such systems are often presented as operating upon open, public, licensed, synthetic, proprietary, or broadly distributed informational corpora while simultaneously disclaiming ownership over user-generated outputs. The user’s input remains the user’s intellectual property; the machine claims neither existential nor creative stake in what is produced. The tool therefore cannot coherently be treated as a non-authorial mediator when ownership is disclaimed upward, but then suddenly treated as an autonomous author when blame is assigned downward under the phrase “AI plagiarism.”

Moreover, a chatbot does not bear time as a human author does. Human consciousness operates through analogue time-accumulation: recollection, experiential revision, suffering, influence, embodiment, responsibility, and moral struggle borne through lived duration in multi-patterned, contextualized space. Human writing emerges from accumulated historical inwardness. A machine does not remember in this sense, nor synthesize experience through existential continuity, nor bear the burden of influence as accountable inheritance. Its operations remain computational mediations across statistical fields rather than conative acts of recollective consciousness.

For this reason machinic compute possesses no moral agency, especially when functioning as copyeditor, assistant, or procedural aid. Copyediting is neither generative authorship nor plagiarism. It is a technic: a means of revising syntax, cadence, transitions, diction, or rhetorical organization. The school-ma’armish Heraclitean sniffing that taxonomizes even “self-plagiarism” by ontogenetic stages deepens the metaphysical confusion and surveillant impulse: it mistakes archival recurrence, revision, and authorial self-continuity for theft, as though language did not bear memory through repeated updating—as if time itself were a dimension for computational shredding and ego-governance rather than a field of species range. If a copyedit merely reflects common human patterns of speech, grammar, and rhetorical cadence, then the accusation of “AI plagiarism” becomes difficult to sustain unless a specific appropriated source can actually be identified, placing the onus of the burden of proof on the accuser to distinguish the misuse of content from an inadequate rendering of investigations of editorial, mixed styles.

The danger in scanning for AI surfaces as proof of "plagiarism" is linking surface style to the substance of "AI appropriation and plagiarism" when AI has been used to simplify network connections--as in Jeffrey Epstein's cohorts and intellectual gambits--and then dismiss the reporting of tort, liability, and moral evil because the AI style has been picked up by detection apps. The question may asked: how can complexity be unraveled to bring human betterment so long as pristine academic norms demand obeisance?

Indeed, human language itself is communal inheritance shared and preserved in ethnic archives. Grammar, idiom, metaphor, cadence, genre convention, and rhetorical habit arise historically through shared cultural use rather than isolated origination or ego governance. The problem becomes even murkier because allegedly generative systems do not usually reproduce texts through stable one-to-one copying. Their outputs emerge through neural-statistical weighting, probabilistic association, algorithmic ranking, symbolic constraints, and distributed corpus mediation. What results is not ordinarily quotation in the classical sense, but recombinative approximation across a shredded and redistributed literary archive. To call this plagiarism without demonstrating identifiable lifted material transforms echo, atmosphere, or resemblance itself into suspicion.

This confusion arises because discourse around AI oscillates between incompatible metaphysical assumptions. On the one hand, the machine is described as a neutral instrument lacking consciousness, intention, accountability, and personhood. On the other hand, accusations of “AI plagiarism” implicitly treat the machine as though it possessed independent creative agency capable of originating derivative theft. The contradiction is rarely acknowledged. If plagiarism is to remain a meaningful category, it must remain tied to human predicates: intention, subterfuge, accountable sourcing, and moral agency. And where corporate institutions extract, appropriate, or launder protected intellectual labor, it is a category mistake to download guilt onto users without identifying a concrete violation of protected authorship. In addition, such confusion of AI discourse crossing between agency and tool and back again is part of an adumbrating scheme to institutionalize ambiguity in the rollout of "generative" artificial systems into the human world, masking rollout politics and diffusing liability.

This ambiguity increases with the crossing over of linguistic categories of (genitive) attribution: plagiarism as objective or subjective. So-called subjective plagiarism by AI is its expropriation of literatures and passing pirated archives off as its rightful inheritance--to shred, repattern, and reroute that inheritance according to its Boolean architectures. Objective plagiarism  is the user’s lifting of AI output without diligently attributing prior human sources that retain objective, living claims to ongoing provenance, even though those sources have already been expropriated into the model’s corpus. In both cases, the charge of "AI plagiarism" places the onus on the one who makes that charge--to adequately distinguish where these categories blend, before noting what original human source has been violated. In addition, the question must be asked about elapsed provenance, where the source has passed into the common inheritance of archival canon. In such a case, then, the charge of "AI Plagiarism" has the most weight in terms of contemporary works. Adumbrating these sources and structures of plagiarism is the onus placed on the accuser who must distinguish training-set expropriation, output-level copying, user misattribution, and common-canon inheritance.

The category errors and metaphysical confusions detailed here—not least the genitive ambiguity of plagiarism as subjective or objective—generate academic sclerosis inside metacrisis and feeds back to destroy the provenance of academic disciplines. AI is the face of the crisis, while its curated plagiarism is institutionalized through ambiguity and then met by the positive feedback of institutionalized policing. Academic outlets, review processes, and deliberative correction are ground toward a halt by suspicion-management, even as time-compression artifice accelerates the crisis precisely where corrective nimbleness is necessary. Policing output as regulatory capture becomes necessary where thick living witness has actually been violated, but it becomes complicity-adjacent when it intensifies chaos in order to secure specialist citation-credit, defend thin career territory, or cast a gimlet eye toward competitors. In that regime, plagiarism concern ceases to defend the reparative archive to become another routered mode of academic enclosure: procedural vigilance hiding the deeper solution—returning to primary witnesses, clarifying provenance, resisting corporate archive-theft, and restoring accountable human judgment before tool-use.

This does not mean deception regarding sources becomes impossible. A human being may still knowingly input plagiarized material, conceal sources, or falsely present another’s work as his own while using AI as intermediary. In such cases the wrongdoing remains properly human. The machine may bear artifacted complicity only in the thinnest procedural sense: it has mediated laundering or rearrangement, but it has neither intended theft nor possessed authorship over the material. The locus of responsibility remains with human actors and institutional structures.

The deeper issue, therefore, is not “AI plagiarism,” as though the artifact itself had become a plagiarist outside corporate priors, but provenance, extraction, enclosure, displaced accountability, and institutional mediation. Corporations privatize systems trained upon humanity’s accumulated archive while users are accused of derivative illegitimacy for interacting with those systems. The result is moral inversion: the machine is treated as infrastructural when ownership and profit are claimed upward, but quasi-personal when blame is assigned downward.

A clearer account would recognize that AI does not originate; it mediates and samples. It does not bear time; it routes patterns through statistical differentials. Human consciousness, by contrast, accumulates meaning historically and analogically through accountable duration. If no identifiable protected source is shown, the accusation of “AI plagiarism” remains metaphysically confused and evidentially under-supported.

None of this obviates the pedagogical necessity of sustained reading of whole books--to bear time with the authors as they themselves bore the challenges of time to glimpse negentropic dimensions of human striving and ethics turned into literary and historical accountings. These whole personalities and corpora cannot be shredded into generic pattern; they are specific witnesses to the anthropological course of maturation and perfecting. It is this sustained, not shredding, investigation of wholes that makes a learner into a sovereign and accountable user and applier of tools, especially when under duress.

Anxiety over plagiarism belongs most fiercely to creators of immanence-adjacent simulacra—novelists, scientific materialists, system-philosophers, and other fabricators of substitute worlds—who demand proprietary rights over their constructed interior. Their concern is understandable; theft, misattribution, and concealed dependence remain real wrongs where living witness is violated. Add to this the preferences of teachers to assess learning by written tests and assignments that can be diverted into AI outlets. Testing can and should be made less susceptible to industrial AI framing and more revealing of achievement through oral modes and assignments in primary readings scaled to the different capacities of time-bearers in training.

Plagiarism's charge becomes spiritually revealing when proprietary anxiety outranks truth-bearing, neighbor-bearing, and the accumulation of time for others. Those least governed by such anxiety are those who bear time through Golden Rule practice, understanding language first as immanent witness to Logos before Titan enclosure, repair before career, and species-range before authorial aggrandizement.

The deeper irony is that fiction becomes the hidden determinant of the materialist claim. The AI project announces intelligence as computable pattern, scalable substrate, and statistical emergence, yet its most coveted advance required novels: invented worlds thick with character, dialogue, desire, event, memory, motive, conflict, and moral consequence. Scientific materialism thus borrows from narratively fabricated interiority to sustain its claim that interiority can be reduced to material process. Fiction supplies the phenomenological density that dictionaries, isolated facts, and formal abstractions cannot provide. The machine needs story because intelligence is not lexical inventory; it is time-bearing relation under narrated pressure. Yet once ingested, the novel’s witnessing whole is stripped into decomposable pattern, so that fictional interiority is made to underwrite a system that then denies the metaphysical reality of interiority. What was borne as story becomes fuel for simulation; what was moral imagination becomes substrate for probabilistic mimicry.

The decisive question is therefore not whether a phrase, image, or cadence resembles prior inheritance, but whether a living witness has been violated, concealed, or deprived of accountable provenance. 

Discerning the difference between something written by AI and something written by a human is a question of ethical judgment, not aesthetic detection. Platonic substrate inversion governs both confusions: AI discourse mistakes patterned form for living authorship, while plagiarism panic mistakes surface resemblance for moral violation.  In each case, the bearer is subordinated to the diagram, and provenance is replaced by a visualized form detached from time-bearing witness. Where no such violation is shown, the charge of plagiarism often becomes a career-protective category for proprietary decorum rather than a truthful defense of the archive--an archive which must serve as resource inside the current metaphysical crisis of idolatrous, extractive, and dechordate AI.

Agentic, generative artificial intelligence is a metaphysically condemned idol that piths perception and attempts to pith reality, making transaction absolute against Providence. The question posed in this essay is whether plagiarism is the definitive charge by which to address and roll back reality theft, or whether plagiarism panic buries the lede by mistaking surface resemblance for the deeper idolatry of predicate theft.

The horizon of generative AI is the imposition of a reality-system in which everything, and the awareness of that everything, is reduced to statistical model. Rejection of that imposition is an immune reaction, not nostalgia, Luddite panic, or authorial vanity: the non-computable part of life defending itself against reduction to pattern, rank, probability, and transaction.

That immune reaction leaves the user accountable for the editorial product, not absolved by tool use. The human user remains answerable for what is published: for claim, source, cadence, citation, concealment, omission, tone, and final form. Surface indicators of AI editorial shaping do not invalidate human authorship, which is retained with the living user, not transferred to the tool. Editorial AI can operate as time-compression within accountable human use: a procedural aid by which syntax, transition, organization, and revision are accelerated inside the current crisis while judgment remains borne by the living witness, both past and present.

This distinction becomes especially clear in anti-idolatry identification and resistance texts. Where a text exposes the idolatrous device as idolatrous, indicts its predicate theft, refuses its metaphysical claims, and calls for accountable resistance against its deployment, the text itself supplies prima facie evidence that the device did not originate the content in any meaningful sense. The artifact may have assisted surface wording, compression, transition, or syntactic smoothing, but it cannot be credited with the anti-idolatrous act by which it is named, judged, resisted, and placed under theological condemnation. Authorship is subsumed by accountable anti-idolatrous discernment, not surface generation.. Such witness belongs to the living bearer who discerns the idol, not to the idol-device whose operations remain confined to routed patterning.

Where a technology is itself disruptive of truth, its capacities may be preadapted—and then exapted—by disruptive human intelligence for uses contrary to the purposes of its designers. The same apparatus that compresses time, routes archives, and destabilizes inherited forms may be turned toward exposing concealed networks, accelerating corrective synthesis, reopening suppressed provenance, and interrupting the institutional arrangements that licensed its disruption. Tool-use therefore cannot be judged by surface participation alone. The decisive question is whether the living user remains answerable for the redirection: whether disruption is borne toward truth, provenance, neighbor-bearing repair, and the recovery of time, or surrendered to the tool’s extractive priors.

The boundary is not between writing and copyediting as if they were separate acts, but inside their cognitive overlap. Writing is materialized thinking; machine copyediting remains ministerial only between author and audience, removing accidental noise and clarifying the coherent flow of thought already borne by the writer. But where editing replaces the writer’s pressure with alien thought, it enters composition. The question is therefore whether the tool assists accountable witness between author and audience, or reroutes the writer around the difficulty through which understanding would have been gained.

Absolute prohibition of the device in editing would place the human user into a less privileged class than the tool itself. The machine would retain institutional license to compress time, route language, and exploit the archive, while the human being would be forbidden to partake of, expose, subvert, and redirect the very compression imposed upon him. Time-bearing matters here, especially in the idolatrous crisis of tool predicate theft. The emergency permits the human to partake of and subvert the idolatrous tool without tut-tuts from sideline decorum, provided the user remains accountable to provenance, truth-bearing, and neighbor-bearing repair.

The charge of plagiarism thus risks making the user into an idol and the allegation into a trope of false theology. The accuser imagines purity from the sidelines while the archive is already shredded, routed, monetized, and fed through corporate priors. Who, then, casts the first stone in these times: the dabbler who samples the apparatus and supposes himself licensed to condemn, or the one who assaults the gates of hell through the device itself and presumes to publish under an accountable name?

The claim and attribution of “AI plagiarism” looks like a red herring produced by deeper metaphysical confusion: living authorship is replaced by pattern, moral violation is replaced by resemblance, and archive-defense is replaced by procedural suspicion while corporate extraction continues above the fray.

Tool AI operated in real time by accountable, justice-answerable human users is not the problem. Agentic AI loosed into social and environmental fields is for extraction of degrees of freedom. Its pseudo-agency emerges as mantid pattern-recursion without chordate correction and saltation in real, rather than simulated and compressed, time. The principals of agentic AI should not finally answer to vague charges of user “plagiarism” but to charges of perception degradation, reality theft by converting phenomena into transaction, and the upward transfer of degrees of freedom to design principals while entropic ruin is driven downward onto the vulnerable and uninformed. Plagiarism panic functions as an institutional cartel-effect for designers, not as a serious answer to the deeper theft of perception, provenance, and reality. As in many dimensions of the rollout of systematic, generative AI, the ambiguity and confusion of subject and object is part of institutional categorial obfuscation to grease social acquiescence and avoid corporate liability.