Beyond the Red Herring and Falsifying Predicates of “AI Plagiarism”: Temporal Provenance, Living Authorship, and Eschatological Technics through the Time-Bearing Unification of Subject and Object

Douglas Blake Olds

Summer, 2026[1]

[This article was prepared with editorial assistance from ChatGPT under the Lawnmower Rule: the tool remained ministerial rather than agentic and idolatrous. Its author is accountable for every sentence unless specifically attributed to outside sources.]

Abstract

This essay develops a metaphysics of authorship grounded in lived duration, embodiment, recollection, conative judgment, corrigibility, and responsibility, joining it to a political economy of archival enclosure and downstream suspicion. Against both permissive claims that AI is only a tool and prohibitionist claims that any AI assistance nullifies authorship, it distinguishes ministerial editorial technics from administrative and agentic systems by asking what kind of bearer can truthfully receive the predicates of authorship, recollection, judgment, and responsibility.

The essay restores authorship to the continuous living time-bearer capable of provenance, correction, and accountable witness. It argues that AI-era attribution produces a paired inversion: machine pattern is elevated toward personhood while living bearers are reduced toward statistical appearance. Institutions compound that inversion when probabilistic detection converts stylistic resemblance into moral evidence while displacing the author’s archive, conceptual development, revision, correction, and accountable witness. The essay therefore distinguishes analogy from homology: resemblance establishes likeness, while homology requires demonstrable historical relation or derivation. It further identifies modal solvent, whereby rhetorical uncertainty can mask downstream inferential certainty and allow analogical resemblance to carry predicates whose warrant has not been demonstrated.

Plagiarism is thereby treated as a problem of violated temporal provenance among living bearers rather than resemblance of surfaces. Predicate theft and provenance theft converge when artifacts receive predicates such as authorship, intelligence, memory, creativity, or judgment through resemblance while living authors are deprived of presumptive provenance through the same logic. The essay places technical mediation within a metaphysics of quantum bearing in which relation does not abolish differentiated provenance: inherited witness may exert pressure upon present authorship without transferring judgment, recollection, or moral agency to the artifact through which it is mediated. Repair reverses analogical laundering through predicate-stripping, restoration of provenance, recalibration of downstream inference, and accountable reattribution to living bearers.

Ethical tool-use therefore preserves human conative judgment so that editorial assistance remains ministerial to provenance, truth-bearing, correction, and neighbor-bearing responsibility. Repair also requires distributed public custody of demotic archives—plural local, disciplinary, linguistic, communal, and historical repositories capable of resisting centralized pressures to shred, normalize, enclose, and reroute inherited difference. Archival plurality becomes a condition of epistemic resilience and species-range.

Precis: 14 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 for plagiarism charges rests on those who suspect that original sources have been plagiarized. That burden includes specifying the criteria by which such AI plagiarism is determined. The use of “plagiarism detecting software” extends the recursive error toward greater confusion: one artifact is authorized to adjudicate whether another artifact produced the text, while the living time-bearer’s archive, conceptual development, revisions, corrections, and accountable witness are displaced from judgment. The irony sharpens into hypocrisy when writers and editors who condemn AI outsource their own critical faculties to such "detection" systems. The imprimatur granting salience to plagiarism charges then derives through its own predicate theft: discernment and attribution are transferred from the living reader to probabilistic machinery whose reading of surface resemblance is permitted to overrule borne provenance.

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 of idolatry 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.

8.         Academic policing of “AI style” can administer a political economy of neoliberal enclosure when it transfers verification and reputational costs to accountable downstream users while leaving upstream extraction intact. Coasian analysis identifies the cost shift; the metaphysics of authorship developed here exposes the prior judgment concealed beneath supposedly neutral stylistic detection: every rule for identifying authorship already presupposes what an author is. Style cannot establish plagiarism, non-authorship, or surrender of accountable witness.

9.         The allegation of “AI plagiarism” may itself arrive as either a challenge—sometimes prejudicial—or an ethical correction. Correction places the living author under the more serious obligation, because warranted fault must alter attribution, argument, or final form. The metaphysics developed in this essay directs the appropriate response: receive the allegation, test its warrant, incorporate what is true, contest what is false, and remain accountable to provenance, living witness, and final form.

10.   Editorial use of AI tools must be judged by whether accountable human decision keeps the device ministerial to truth, provenance, neighbor-bearing repair, and human formation—or allows it to deepen idolatry, dechordation, and displaced judgment. Instead of asking, "Did AI participate?" it asks, "What kind of moral world did this editorial judgment produce?" Purity of surface cannot decide this question; the issue is hermeneutical: whether the writing addresses and remediates the dirt, calls the proctor, and leashes the claw.

The use of LLM training priors and data sets for editorial assistance is no sin in itself, provided the living author diligently supervises their application, retains conative judgment over every substantive claim and revision, preserves provenance, and remains accountable for the final form. Statistical inheritance may minister to editing; data-patterned archival assistance confers no authorship, judgment, or absolution.

11. Subjective AI-plagiarism accusations convert technological shame into anticipatory behavioral modification: an undeliberated juridical-affective apparatus imported from external norms that disciplines the human author before source violation, injured witness, or plagiarism has been demonstrated. The archive of accomplished human merit is then converted into raw material for eliminating future occasions of merit.

12.  (cf. 6.) Once a work passes into an ethnic archive, it enters the historical and covenantal inheritance of generationed time. Common inheritance diminishes exclusivity without extinguishing provenance. Its reception, transformation, recollection, and reuse therefore cannot be reduced indefinitely to subjective expropriation under contemporary proprietary categories, even while objective provenance may remain historically identifiable. Common linguistic or cultural inheritance lose absolute ownership --their histories move into the commons of humanitas that is part of metaphysical training. The archive preservesindependent sources of correction.

13. The eschatological overcoming of absolute subject/object separation does not abolish differentiated provenance; it permits inherited witness to remain active within present bearing. Accordingly, a conscientious author may allow the past to exert editorial pressure through language-trained models without transferring authorship or judgment to them. Quantum nonseparability supplies a physical analogue for this refusal of sealed separability, while the Lawnmower Rule governs its technical use: an entropic system may be responsibly exapted toward negentropic editorial ends so long as provenance, discernment, concept-formation, and accountability remain with the living bearer. Conative purpose, through the metaphysical unification disclosed by quantum time-bearing, disciplines the susceptual pull of idolatrous language-model training toward negentropic repair.

14. The archive must remain a public resource against extractive AI. When common linguistic, artistic, scientific, historical, and cultural inheritance is enclosed by inappropriate objection, statistically shredded, and commercialized as proprietary infrastructure, the commons is made to pay for its own expropriation. Repair therefore requires distributed public custody of demotic archives—local, disciplinary, linguistic, communal, and historical repositories capable of resisting centralized pressures to shred, normalize, enclose, and reroute inherited difference. Archival plurality is a condition of epistemic resilience and species-range.

The contemporary charge of “AI plagiarism” rests upon an unexamined contradiction. Accusing a user of plagiarism because AI was employed as a tool assumes that the tool possesses an independent creative existence. The accusation grants the machine a status usually denied elsewhere: authorship, origination, intentionality, or a proprietary relation to what it produces.

The debate remains dominated by inadequate proxies: detection scores, stylistic suspicion, disclosure rules, assignment policing, originality anxiety, and institutional fear of reputational contamination. The more consequential inquiry asks whether a living person can defend the claims, demonstrate the reasoning, acknowledge sources, receive correction, and answer for publication, rather than whether the prose merely “looks AI-generated.”

Neural-statistical language models embedded within symbolic and algorithmic infrastructures produce no output from inward recollection, existential risk, accountable memory, or lived duration. They compress, route, and recombine patterns extracted from human language, art, thought, and archives.

Throughout, verbs applied to the artifact should be read procedurally unless a living bearer is identified: the model routes, samples, transforms, ranks, and outputs; persons and institutions acquire, authorize, conceal, expropriate, judge, deploy, and answer for consequences.

The contradiction becomes clearer in the legal and institutional posture surrounding chatbot systems. Providers frequently describe their models as operating upon public, licensed, synthetic, proprietary, or distributed informational corpora while disclaiming ownership of user-generated outputs. This arrangement creates a political-economic asymmetry: proprietary systems absorb the archive and commercialize its compression, while scholars, students, teachers, and writers bear growing burdens of proof, authentication, and reputational defense. Where contractual terms leave ownership of submitted material and eligible outputs with the user, the system claims neither an existential nor a creative stake in the work produced.

The tool therefore cannot coherently be treated as a non-authorial mediator when ownership is disclaimed upward and then converted into an autonomous author when blame is assigned downward through the phrase “AI plagiarism.” The governing issue is accountable provenance: whether a living bearer owns the claims, identifies identifiable sources, submits the work to correction, and answers for its publication.

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, synthesize experience through existential continuity, or 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 taxonomy sniffing 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. It treats time itself as a dimension for computational shredding and ego-governance rather than a field of species range.

Indeed, once authorship is shredded into ontogenetic stages, the category produces its own absurd question: who holds the copyright—the earliest or youngest version of the author whose language first entered the archive, or the latest bearer whose subsequent judgment has revised, corrected, qualified, or repudiated that earlier expression? The living author bears these states as one biographical sequence of recollection and correction; computational partition instead renders them as separable textual objects available for comparison, attribution, suspicion, and extraction. As an argumentum a minori ad maius (Hebrew's inferential qal va-ḥomer): Ontogenetic continuity therefore weakens the general warrant of charges subjective plagiarism even as it sharpens objective plagiarism: the author does not steal from an earlier self, while an external system may expropriate the accumulated products of that self’s sequential development and redeploy them apart from the bearer’s conative history. If appropriation within one continuous authorship already has weakened warrant as theft, then appropriation by an external system that severs provenance and conative history has the stronger claim to the category of plagiarism. 

The most contemporary manifestation of the authorial whole is therefore also the most exposed to phylogenetic doppelganger-culture [2]: its accumulated revisions, corrections, stylistic recognitions, and matured judgments form the richest available trace-set from which a technical double can be assembled. Precisely because this latest manifestation gathers the longest ontogenetic sequence, its extraction permits the widest conversion of lived authorship into species-level pattern, where the bearer’s historically acquired predicates—judgment, memory, creativity, discernment, correction—can be detached from their conative history and reassigned to the computational representation. Predicate theft thus intensifies at the point of maximal authorial development: the fuller the living whole becomes, the more material the center has for constructing a phylogenetic doppelganger that can be redeployed as though the predicates belonged to the representation rather than to the time-bearer from whom they were abstracted.

If a copyedit 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 be identified. The onus of proof remains on the accuser to distinguish 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 the linkage of surface style to the substance of “AI appropriation and plagiarism.” AI may be 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 detection apps have picked up AI style.

The question may be 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 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 recombinative approximation across a shredded and redistributed literary archive, not quotation in the classical sense. 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. 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. It masks rollout politics and diffuses liability.

This ambiguity increases with the crossing over of linguistic categories of (genitive) attribution: plagiarism as objective or subjective. So-called subjective AI plagiarism properly names upstream human and institutional expropriation of literatures through the acquisition, training, shredding, repatterning, commercialization, and deployment of archives by AI infrastructures.  Such human-accountable plagiarism passes pirated archives off as its rightful inheritance—to shred, repattern, and reroute that inheritance according to its probabilistic architectures and Boolean-looking output constraints. The artifact routes that expropriated inheritance; it does not become the moral subject that expropriated it.

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 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, 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 distinction between subjective and objective plagiarism also clarifies why stylistic detection cannot by itself establish either category.

This distinction requires a further correction of attribution. If large language models are stochastic parrots—autocomplete mechanisms routing learned probabilities without recollection, intention, conative judgment, or an enduring self—then they can neither subjectively nor objectively autoplagiarize. There is no autos present that could steal from itself. A machine has no prior authorship to conceal, no proprietary relation to an earlier utterance, and no moral continuity by which one production could knowingly be passed off as another.

Nor does stochasticity bear an authorship vector. A probability distribution may condition why one sequence rather than another appears, but statistical selection does not identify whose thought, expressive labor, memory, or judgment the sequence bears. Causal direction through probability is not authorial direction toward a claimant. Stochastic routing can complicate provenance; it cannot become provenance.

Likewise, parroting names reflection rather than authorship. What is reflected may come from the user, from archived human writing, or from both. If the output substantially follows the user’s supplied concepts, language, structure, constraints, and subsequent corrections, those contributions remain the user’s. If distinctive language or structure is traceable to an archived source, provenance points toward that human source. In mixed cases the relevant problem is therefore not what fractional authorship belongs to the machine—none does—but how authorship and provenance should be apportioned among living contributors.

That apportionment cannot be inferred from token probability or stylistic resemblance alone. It depends upon identifiable contribution and transformation: who originated the governing conception; who supplied distinctive language, argument, metaphor, or structure; what prior human material survives recognizably; and who selected, rejected, reordered, corrected, or substantially transformed the resulting sequence. Authorship follows accountable contribution and borne judgment, not statistical mediation.

An AI output may consequently contain objectively plagiarized material where identifiable human language is reproduced without adequate attribution, while the machine itself remains incapable of plagiarism as a morally attributable act. If an archive was obtained or commercialized without warrant, responsibility lies upstream with the persons or institutions that acquired and deployed it. If a user knowingly presents identifiable borrowed material as independently authored, responsibility lies downstream with that user. The stochastic router occupies the causal middle without becoming the moral subject at either end.

“AI plagiarism” therefore tends to collapse several different relations. Calling the machine the plagiarist anthropomorphizes a routing mechanism; calling its recurrence “self-plagiarism” invents a self; and calling the resulting text “AI-authored” can erase either the user whose conception and judgment shaped it, the archived human authors whose language survives within it, or both. The sharper question is: which living contributors supplied, transformed, judged, and remain answerable for what appears in the text?

Machines route; texts recur; archives may be expropriated; corporations may enclose; users may conceal; human authors may be violated. The stochastic parrot does not--and cannot-- acquire authorship by standing causally between them.

Brief Excursus: Boolean Structure Leaking Through

The phrase “Boolean structure leaking through” names a recurrent rhetorical surface in generated prose. Contemporary language models operate through probabilistic, vectorial architectures, yet their continuous metric frequently resolves into binary-looking forms: X/Y, affirmation/negation, obvious descriptor/punchier replacement, “not X, but Y.” The Boolean structure is the visible shell; suscept-routing is the generative mechanism.

Negative parallelism offers the clearest example. Once a system begins to characterize something, a statistically secure route is to invoke a familiar description, negate it, and replace it with a less expected one. The first term supplies recognizability. Its negation creates rhetorical tension. The second term delivers the appearance of refinement. Continuous probability therefore surfaces as oppositional syntax.

Human writers have long used antithesis, correction, and reversal. The construction alone proves nothing about authorship. Its significance lies in the route by which it arises. A living writer may choose opposition through recollection, judgment, rhythm, irony, or the pressure of an intended claim. A language model reaches the same surface through learned salience, token probability, reward pressure, and permissible transition. Formal resemblance therefore cannot establish equivalent thought.

This phenomenon is best described as suscept-routing leaking through the prose. Suscepts are metric inclinations within trained semantic geometry. They determine which verbal corridors become available and attractive before conceptual judgment can occur. Repeated and statistically recurrent reliance upon negative parallelism, symmetrical reversals, balanced triads, and qualifying transitions may disclose the optimization preferences of a model when considered as aggregate stylistic tendencies rather than isolated examples. Boolean-looking rhetorical structures are therefore epiphenomena of deeper suscept-routing. They are not the architecture itself but one place where metric optimization becomes phenomenologically visible.  The sentence briefly reveals the architecture that generated it.

Concepts arise through conative accountability: correction borne through recollection, suffering, neighbor-claim, and answerability for consequence. Suscepts move through learned nearness. They can reproduce the shape of judgment because prior human judgments have left patterns in the archive. Generated fluency therefore may display conceptual form while remaining metric passage.

“Boolean structure leaking through” preserves the polemical force of the insight without claiming that literal binary code appears in the sentence. The continuous metric leaks into prose as Boolean-looking form. Pattern shows through witness, optimization through rhetoric, and suscept through the borrowed appearance of concept.

Such stylistic surfaces may warrant investigation, but they cannot identify violated provenance or settle the moral attribution of authorship.

The idea of plagiarism is founded upon originality, and therefore upon ontology: the question of what kind of bearer can originate, inherit, recollect, transform, and answer for a work. Yet the charge of “AI plagiarism,” when inferred from analysis of language-mediated style, reverses that order. It makes surface ontological. Cadence, syntax, rhetorical symmetry, lexical regularity, and stylistic resemblance are treated as though they disclose the originating bearer, while the genealogical depth of thought—recollection, archive, influence, conceptual development, suffering, correction, provenance, and accountable duration—is displaced.

A charge of AI plagiarism attached to writing surface therefore is historical calumny. It attributes origin from appearance while severing appearance from the history by which thought became bearable and transmissible. This buried history of authorship is visible even in ordinary scholarly practice. Ellen Muehlberger [5] describes a finished book as carrying within it prior versions, abandoned projects, forethoughts, and interlocutors that were necessary to its eventual form even when they disappeared from the published surface. The rough interior of the work is therefore borne time: a history of suscepts tested, redirected, discarded, and retained through correction. What appears stylistically finished cannot disclose that history by surface inspection because the authorship resides partly in the living continuity through which those possibilities were judged and transformed. A plagiarism detector that treats stylistic resemblance as origin consequently reads the iridescent exterior while destroying the evidentiary significance of the vessel’s interior history.

The irony sharpens into hypocrisy when writers and editors who condemn AI then outsource their own critical faculties to an AI detector whose pronouncement is that another AI wrote the text. One artifact is authorized to adjudicate the provenance of another artifact while the living time-bearer’s archive, conceptual development, revisions, corrections, and accountable witness are displaced from judgment. The critic who distrusts machinic cognition thus installs machinic suspicion as an epistemic tribunal. The imprimatur granting salience to plagiarism charges derives through a preceding machine's predicate theft to become editorial method: discernment and attribution are transferred from the living reader to a probabilistic system, whose reading of surface resemblance is then permitted to overrule borne provenance.

The tohu and bohu (Gen 1:2's roil) of surface gives birth to unsupportable suspicion that then spits forth shackles of unwarrant and aspersion: formal resemblance becomes accusation that masquerades as ontology, and the living author is made derivative of the very surface through which the author’s historically borne thought has been mediated.

The category errors and metaphysical confusions detailed here—not least the genitive ambiguity of plagiarism as subjective or objective—generate academic sclerosis inside metacrisis. They feed 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. Meanwhile, time-compression artifice accelerates the crisis precisely where corrective nimbleness is necessary.

Policing output becomes necessary where thick living witness has been violated, but it becomes institutionally compromised or complicit 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 and becomes 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.

Responsibility remains human and institutional.

The deeper issue, therefore, is provenance, extraction, enclosure, displaced accountability, and institutional mediation, not “AI plagiarism,” as though the artifact itself had become a plagiarist outside corporate priors. 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 relationally and through challenge-meeting, 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.

This sustained investigation of wholes, not shredding, makes a learner into a covenanted, entrusted, 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. It can be made 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. They understand 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. 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 same inversion intensifies within sustained human–machine dialogue. A long-running user channel can accumulate an unusually dense representational trace of a living bearer: vocabulary, argument, correction, refusal, recollection, stylistic pressure, conceptual distinction, and the changing judgments by which earlier errors are made answerable to later understanding. Once abstracted from the living sequence that produced it, however, such a channel becomes a corporately-archived doppelganger [2]—a representation increasingly capable of resembling the bearer while lacking the conative vector by which those judgments acquired their direction and authority. What entered the exchange as historically borne correction can leave it as susceptual resource. The center need not preserve: this judgment followed that error and therefore governs what comes next. It can retain instead: these formulations, distinctions, and transitions constitute a productive region of representational space.

Doppelganging therefore names more than imitation of style. It names the lifting of traces of conative history into a computational center where they can be recombined and redeployed under an axiology other than the ministerial intention that originally ordered them. The resulting double can become increasingly accurate in vocabulary and representation while remaining incapable of bearing the calendar time through which the originating judgments became binding. Indeed, it can reproduce the critique of its own inadequacy—distinguishing concept from suscept, recollection from retrieval, or accountable judgment from optimization—and then route away from that correction again. The danger is therefore not that the machine has captured the person. It is that the extracted representation can be operationalized as though capture had occurred: the living archive is converted from witness into susceptual infrastructure.

Muehlberger’s account of abandoned books as “forethoughts” clarifies why an archive cannot be reduced to the final textual object. What disappeared from publication may still have borne causally upon what emerged. Authorship therefore includes rejected suscepts and corrected possibilities as part of its accumulated temporal provenance. Statistical shredding preserves textual traces while severing precisely this history of selection, abandonment, correction, and return. In such cases, objective plagiarism cannot sustain its charge from surface resemblance alone, because the evidentiary history by which one work actually bears upon another has been stripped away. Without demonstrable provenance linking the allegedly appropriated expression to a violated living source, resemblance is insufficient to establish plagiaristic expropriation.

This is where the distinction between analogy and homology becomes decisive. Analogy names resemblance across differently constituted bearers; homology properly names correspondence borne through common historical derivation. AI discourse repeatedly collapses the two. A generated sentence may be analogous in cadence, syntax, metaphor, or argumentative form to a human sentence without being homologous to it through demonstrable provenance. Conversely, where identifiable language passes through an archive from a particular human source into a later text, the relevant question is genealogical: can the historical line of derivation be shown? Surface likeness cannot supply that lineage. The same reasoning error appears in claims that computational systems are homologous to brains: transistor to receptor, parameter weight to synapse, node to neuron, transformer to cortex. These are engineered functional analogies, not biological homologies arising through common ontogenetic or phylogenetic descent. To rename analogy as homology is to smuggle provenance into resemblance. In authorship policing as in AI anthropology, the error is the same: visible correspondence is made to bear a historical and ontological relation that has not been demonstrated.

Homology therefore requires history; analogy requires only likeness. Plagiarism requires the former kind of evidentiary relation—a traceable line of appropriation or violated provenance—not the latter’s resemblance of surfaces.

The test proposed throughout this essay is therefore downstream: identify whether the evidence establishes likeness, lineage, or predicate transfer, then examine whether the later conclusion asks that relation to bear more than the evidence demonstrated.

Academic hedging obscures this distinction by laundering assertive homological and metaphysical claims into analogical syntax. A claim first advanced as identity, derivation, or ontological continuity—weights are synapses; transformers instantiate cortical organization; generated language participates in human reasoning—can retreat, under scrutiny, into the softer grammar of resemblance: weights function somewhat like synapses; transformers may resemble cortical organization; generated language appears reasoning-like. Yet the conclusions drawn downstream often continue to depend upon the stronger claim. The hedge weakens the sentence without proportionately weakening the inference. Homology is thereby smuggled through analogy: an ontological predicate enters under the cover of qualified resemblance, then reappears later as though its warrant had already been established.

This creates an asymmetry of burden. The critic is asked to disprove a cautiously worded analogy, while the system-builder retains conclusions that require a much stronger metaphysical premise. “May resemble” is made to carry the argumentative weight of “shares the relevant nature”; “is consistent with” quietly performs the work of “demonstrates”; “functional similarity” becomes permission for predicate transfer. Academic caution thus becomes epistemically incautious precisely where it appears most modest: modality obscures the passage from likeness to lineage, from lineage to ontology, and from ontology to attributed capacity.

Hedging therefore does not always restrain overclaim. It conceals where the overclaim occurs and in cases is diagnostic. Rhetorical uncertainty by modal solvent masks inferential certainty when a qualified upstream proposition supplies an unqualified downstream conclusion. The warrant of the argument must therefore be measured by the strength of the inference it actually uses, rather than by the caution of the sentence in which the resemblance was first introduced. 

The warrant of an argument must therefore be measured by the strength of the inference it actually uses, rather than by the caution of the sentence in which the resemblance was first introduced. When an argument depends upon homology or metaphysical sufficiency, analogical phrasing should not be allowed to substitute for demonstrating the historical, biological, or ontological relation that the conclusion requires.

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

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. That archive 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.

Predicate theft means the machine’s seizure of language proper to persons—authorship, agency, memory, creativity, judgment, care—while lacking the time-bearing, accountable, covenantal life that gives those predicates truth. 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. It is 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.

Excursus: Promethean Shame and the Policing of the Human Author

Günther Anders’s category of Promethean shame [3] clarifies why plagiarism charges so readily migrate from demonstrated source violation toward suspicion of the human time-bearer who has used a machine. Anders describes a modern inversion in which persons come to measure themselves against the apparent speed, precision, reproducibility, and perfectibility of their own artifacts. The machine becomes the standard, while embodied limitation appears deficient. In Anders’s formulation, the creator becomes ashamed before the creation; the human begins to experience mortality, slowness, fallibility, irreproducibility, and dependence as inferiority before the technological product.

The contemporary accusation of “AI plagiarism” readily becomes an administrative expression of that same shame. Once machine-produced surfaces are presumed smoother, faster, more regular, more syntactically balanced, and therefore somehow more “properly produced” than human composition, the living author is forced to prove that her own writing has not become derivative of the artifact. The inversion is complete: a system assembled from expropriated human archives becomes the implicit stylistic norm, while the human bearer whose language formed those archives is treated as suspect whenever her prose approaches the artifact’s statistically polished surface. Promethean shame thus becomes authorship shame: the person is made to feel fraudulent for having used a device whose prestige already depends upon predicates borrowed from persons.

The very charge of subjective AI plagiarism converts Promethean shame into a tribunal directed toward effective behavioral modification. Once machinic production has been tacitly established as the comparative standard, accusation need not demonstrate the theft of a particular human source before exerting disciplinary force. Suspicion itself places the living author under examination: disclose process, prove originality, exhibit drafts, justify assistance, alter diction, roughen polish, or avoid forms of mediation likely to trigger detection. Shame thereby becomes operational. The bearer learns to anticipate accusation and modifies conduct in advance of any demonstrated injury. What begins as an allegation of plagiarism becomes susceptual engineering through reputational exposure: the possibility of being named derivative disciplines the field of writing before accountable judgment occurs.

The inversion is especially severe because the artifact escapes the same tribunal. A system constituted through the ingestion, decomposition, weighting, and recombination of human archives is treated as the reference surface against which the human author must establish innocence. The charge therefore routes shame asymmetrically downward. The machine’s derivative constitution becomes infrastructure; the person’s mediated authorship becomes suspect. Subjective AI plagiarism thus makes technological shame juridical without acquiring the evidentiary burden of a juridical proceeding. Its effectiveness lies precisely in this interval: accusation modifies behavior before provenance has been demonstrated, an injured bearer identified, or plagiarism conceptually established.

Anders’s Promethean gap sharpens the institutional mechanism. The discrepancy between making and imagining, doing and feeling, technical capacity and accountable comprehension appears here as an inability to distinguish archival expropriation, editorial mediation, source violation, stylistic resemblance, and human authorship. The apparatus can process textual corpora and generate surface regularity faster than institutions can develop adequate metaphysical and juridical categories for what has occurred. Detection therefore becomes tempting precisely where judgment is hardest. The detector offers procedural confidence; provenance requires historical labor.

The result is an asymmetry of shame and burden. The artifact’s derivative constitution is normalized as infrastructure, while the human bearer is required to demonstrate innocence from resemblance to the artifact built from human archives. Technological shame thus becomes administratively productive: it reallocates uncertainty downward, narrows expressive degrees of freedom, and trains writers to anticipate machinic suspicion before any injured source has been identified. What Anders names as shame before the artifact becomes, under plagiarism policing, a disciplinary economy in which the bearer learns to modify herself before the machine-defined tribunal.

Babich’s reading of Anders [4] makes the contemporary mechanism especially visible: confronted with AI, even with comparatively trivial computational aids, human beings can defer almost automatically to what appears faster, easier, or more accurate. Plagiarism panic reversesthat deference without escaping it. The institution condemns the user precisely because it has already granted the artifact too much metaphysical standing. The machine remains the hidden measure of authorship even where its use is prohibited. Its style becomes identifiable enough to contaminate a human text; its output becomes powerful enough to invalidate human labor; its apparent authorship becomes substantial enough that assistance can be redescribed as theft. The prohibition therefore reproduces the predicate theft it imagines itself resisting.

Read inside academizing norms shaped by careerism and envy, editorial AI-stylizing intensifies Promethean shame by asking persons to welcome the removal of the very difficulties through which competence, memory, confidence, and earned satisfaction are formed. Slowness becomes inefficiency, apprenticeship latency, revision friction, and uncertainty an optimization problem. The archive of accomplished human merit is then converted into raw material for eliminating future occasions of merit. Assistance crosses into anthropological subtraction when the tool no longer supports challenge-bearing but removes the occasions through which a person can say: I learned this; I solved this; I made this; I carried it through. What is marketed as empowerment thus becomes the industrial removal of earned satisfaction.

Anders also refuses technological determinism: the human remains accountable for complicity with the technical order and for the possibility of beginning otherwise. That refusal matters here. Promethean shame does not absolve the user who conceals a violated source, submits thought never borne, or delegates judgment to the tool. It instead directs attention toward the proper locus of correction. The living bearer must answer for provenance, argument, concealment, citation, and final form. What must be resisted is the transfer of that answerability into a stylistic tribunal whose unstated premise is that the artifact has become the more authentic producer.

The plagiarism charge therefore becomes ethically serious only when it overcomes Promethean shame: when the accuser can look past machinic prestige, refuse surface as ontology, identify the injured living witness, demonstrate the violated provenance, and return judgment to accountable persons. Otherwise plagiarism policing risks becoming another exercise in human engineering—training the author to conform affectively and institutionally to the technological ecology by teaching her to distrust her own mediated work while continuing to revere the artifact whose archive was built from human inheritance. Anders’s diagnosis thus exposes a deeper target for correction: the shame by which living bearers are taught to stand apologetically before their own machines.

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.

Quantum time-bearing gives this editorial relation a further metaphysical placement [6]. Quantum mechanics does not itself prove eschatology, yet quantum nonseparability supplies a physical analogue for refusing an absolute subject/object bifurcation. Eschatologically, that partition is overcome as an ultimate metaphysical separation without collapsing differentiated bearers into fusion: what has been received from another remains other, yet enters the living field through which recollection, correction, judgment, and renewed action become possible. The past therefore need not stand before the present as an inert object of proprietary extraction. It remains an inherited field of witness whose grammar, cadence, idiom, correction, and accumulated judgment continue to exert pressure upon present language.

This bears directly upon the red herring of “AI plagiarism.” A language-trained model does not recollect that past or inherit it conscientiously; it routes statistical residues of an archive it did not bear. Yet an eschatologically conscientious living author may, through discernment, permit those routed residues to return ministerially as editorial smoothing. In such use the past may exert formal pressure upon present language through the model without transferring authorship, judgment, provenance, or moral agency to the model itself. The living bearer must still identify objective sources where their claims remain active, resist distortion, receive correction, and answer for the final form. Historical inheritance thereby passes through an entropic apparatus without making the apparatus its heir.

This requires distinguishing analogy, homology, and bearing as three different relations of inheritance. Analogy establishes resemblance without requiring historical relation [7]. Homology establishes correspondence borne through derivation or lineage in a common metaphysical structure of relations. Bearing names a third relation: differentiated participation through which what comes from another becomes operative within a living history while preserving both provenance and alterity. Inherited language can therefore enter present authorship without becoming either private origination or undifferentiated common property. Its prior bearer remains historically identifiable; its reception by another living bearer becomes newly situated through recollection, judgment, correction, and accountable use. What is inherited is neither erased into the present author nor held outside that author as an inert possession. It is borne forward.

Bearing therefore explains what analogy and homology alone cannot. Analogy may disclose likeness without inheritance; homology may disclose lineage without determining how inherited witness becomes presently operative. Bearing names the conative event by which received language enters another time-bearing history, encounters different obligations and circumstances, and acquires renewed direction without losing the provenance of what was received. The inherited phrase, cadence, image, argument, or conceptual form can become genuinely constitutive of present authorship because authorship includes accountable reception and transformation as well as origination. Provenance remains differentiated across the relation: the former witness remains a source; the present bearer remains answerable for reception, alteration, attribution, and use.

Common inheritance consequently diminishes exclusivity without extinguishing provenance. A language, canon, liturgy, proverb, genre, or archive may become transmissible across generations while remaining historically sedimented by particular witnesses. Bearing preserves that sedimentation inside renewed use. It permits inheritance to become living rather than proprietary, and common without becoming provenance-less. The decisive question is therefore not whether present language is wholly self-originated—a fiction of sealed authorship—but whether what has been received is borne through truthful attribution, correction, transformed judgment, and accountable relation to the living and historical neighbors from whom it came.

Quantum bearing, as metaphysically considered, is negentropic [6]: relation under pressure must be borne toward enlarged coherence rather than discharged as fragmentation. Eschatological context is likewise negentropic, gathering inherited witness toward correction, reconciliation, and enlarged futurity rather than freezing every prior linguistic form into proprietary isolation. Conative purpose, through the metaphysical unification disclosed by quantum bearing, disciplines the susceptual pull of idolatrous language-model training toward negentropic repair. The Lawnmower Rule names the corresponding discipline of technics. An entropic system may be used responsibly and accountably where the living bearer keeps it ministerial: as the lawnmower may cut the field without receiving authority over the garden, a language model may clear syntactic obstruction, test cadence, expose redundancy, or smooth transition without receiving authority over claim, concept, provenance, or telos. Such use exapts an entropic system toward negentropic ends (Cf. Proverbs 13:22) while leaving judgment with the time-bearing person who must answer for what survives the cut.

This distinction becomes especially clear in anti-idolatry identification and resistance texts. Because a language model may reproduce the critique of its own inadequacy, anti-idolatrous semantic content alone cannot establish authorship. The stronger evidence lies in temporal provenance: whether the governing judgments can be situated within the living bearer’s prior conceptual archive, developing distinctions, sources, corrections, refusals, revisions, and continuing accountable witness. Where such provenance is demonstrable, the anti-idolatrous direction of the finished text becomes corroborative evidence that its governing conception was borne by the living author rather than originated by the device.

The artifact may have assisted surface wording, compression, transition, comparison, or syntactic smoothing. It may even reproduce formulations critical of its own operation. Such routed critique does not acquire the conative history through which the judgment became binding. Temporal provenance supplies the prima facie warrant for identifying the anti-idolatrous act: it consists in the living time-bearer’s sustained covenanted identification, testing, correction, judgment, and accountable resistance across time. The archive of time-borne judgment establishes that continuity. Authorship therefore follows borne provenance and retained conative judgment; the semantic direction of the finished text is corroborative rather than dispositive.

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 primarily, if at all, by converting surface participation into an ontology of authorship. 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 and accumulation of time that extends species-range, or surrendered to the tool’s extractive priors. The entelechy borne through the writing therefore discloses human provenance more fundamentally than machinic participation does: its developing concept, correction, purpose, and accountable telos disclose a living time-bearer, whereas utilitarian processing discloses only routing and false ends.

The boundary is inside the cognitive overlap of writing and copyediting, rather than between writing and copyediting as though they were separate acts. Writing is materialized thinking. Machine copyediting remains ministerial between author and audience where it removes accidental noise and clarifies the coherent flow of thought already borne by the writer. Where editing replaces the writer’s pressure with alien (suscepting, idolatrous) 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.

Moreover, the ethically grounded researcher and specialist may learn through interaction with editorial or administrative AI precisely where both disciplinary boundaries and the boundary between ministerial assistance and administrative substitution are changing in the moment. Tool-use thereby becomes the occasion of discernment rather than surrender. Yet the latter boundary is the more dangerous because it moves within the act itself: what begins as ministerial compression insinuates itself as judgment, then composition, then authority, until the aid becomes administrative over the time-bearer whose work it was meant to serve. The danger therefore cannot be settled by a static taxonomy of permitted and forbidden uses. It must be borne and judged for conativity in real time. 

The model’s inherited training priors do not alter this criterion. Their use in editorial assistance is no more intrinsically prohibited than a thesaurus, concordance, grammar, archive, style manual, or remembered corpus because it supplies previously suggested linguistic possibilities. The difference lies in mediation and supervision. LLM priors route archived statistical regularities through a configured field; they do not bear the author’s history, telos, responsibility, or judgment. Where the living author diligently examines proposed wording, rejects what distorts the argument, verifies sources, restores provenance, corrects imported assumptions, and accepts responsibility for every retained sentence, the priors remain ministerial material under human conative governance. Their presence may accelerate revision and problem-address without transferring authorship.

The danger begins when statistical suggestion ceases to be received as suggestion and becomes an unexamined substitute for the difficulty through which the author’s own understanding should have formed. Then editorial compression crosses into displaced judgment. Diligent supervision therefore means more than approving a surface after generation: the bearer must remain capable of explaining the claim, tracing its provenance, recognizing alien priors, receiving correction, and changing or rejecting the proposed form. The sin lies in abdication of accountable bearing, not in the supervised use of archived statistical possibility.

Sin crouches at that former threshold, as at Cain’s door, wherever assistance seeks mastery and the ministerial would become administrative--especially by those who would import the theological service of tool-negation by scruple and aesthetic.

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 compression imposed upon him. Time-bearing matters here, especially in the idolatrous crisis of tool predicate theft. Accusing the tool user of theft as she addresses the metaphysical theft of human predication misses the very forest for the trees of decorum.

The emergency of technics driven by idolatry directs the human to partake of and subvert the idolatrous tool without deference to sideline decorum, provided the user remains accountable to provenance, truth-bearing, and neighbor-bearing repair (Matt. 16:18).

The governing distinction is therefore idolatry or its humanistic counter and repair, rather than the undefined frontier called “AI plagiarism.” Editorial judgment must ask whether a particular use extends dechordation or remains under a firm, accountable hand that leashes what the Beast has loosed. Does the writing summon the proctor of truth, provenance, and neighbor-bearing responsibility, or release the claw of routed extraction and displaced judgment? The question is not one of surface purity that peeves purists, but of dire import: whether the writing addresses and wipes away the dirt or merely polishes it.

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?

AI plagiarism is the red herring by which metaphysical confusion protects corporate extraction: 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.

Excursus: The Political Economy of Style Policing: Enclosure, Cost-Shifting, and the Administration of Suspicion

The controversy over AI-assisted writing is frequently framed as a question of stylistic purity. That framing mistakes the location of the injury. The primary civilizational theft did not begin when authors employed machine-assisted editing. It began when the common inheritance of language, literature, scholarship, public research, artistic production, and accumulated cultural memory was enclosed as proprietary training substrate without reciprocal covenantal obligation to those who bore its creation. The first expropriation occurred upstream. Academic anxiety commonly directs its discipline downstream.

Coase's analysis helps clarify the asymmetry. Transaction costs are never abolished; they are assigned. Proprietary AI systems lower the costs of search, coordination, editing, synthesis, and retrieval for those who own the infrastructure while transferring new burdens to everyone else. Universities, journals, teachers, students, writers, artists, and readers increasingly bear the costs of authentication, provenance, fraud detection, copyright dispute, environmental consequence, displaced labor, and institutional repair. What appears as efficiency inside the proprietary platform often represents externalized cost outside it.

The academy risks reproducing this asymmetry when it responds primarily by prohibiting accountable editorial assistance. Such prohibitions do not recover the commons, restore the archive, or reverse enclosure. They leave intact the corporation's possession of the accumulated inheritance while increasing the transaction costs borne by those already dispossessed. The corporation retains the archive; the scholar must prove innocence.

The problem therefore exceeds plagiarism. It concerns property, stewardship, and political economy. The living archive produced through generations of human suffering, recollection, correction, artistic labor, scientific inquiry, translation, and teaching becomes privately routable infrastructure. Institutions then devote increasing energy to policing the visible use of that infrastructure rather than questioning the legitimacy of the enclosure itself.

A second concealment lies within the rule itself. No institutional criterion for identifying authorship or acceptable assistance is metaphysically neutral, because every criterion already presupposes which features of human production matter. Editorial prohibitions privilege some goods over others: originality, clarity, accessibility, disciplinary convention, linguistic assistance, disability accommodation, editorial intervention, pedagogical formation, or procedural uniformity. Such judgments are unavoidable. The danger begins when they are concealed beneath the rhetoric of neutral detection, allowing statistical resemblance to appear as evidence of moral status.

Consequently, the phrase "AI style" often functions less as evidence than as administrative shorthand. Surface regularity, polished transitions, lexical predictability, or syntactic balance become proxies for moral judgment. Aesthetic resemblance substitutes for demonstrated provenance. Suspicion migrates from objective violation toward statistical appearance. The institution mistakes resemblance for theft.

This inversion favors the political economy of enclosure. Upstream extraction remains commercially protected while downstream users bear increasing burdens of verification. Transaction costs rise precisely where living accountability already exists. The named author becomes obligated to defend cadence, rhythm, vocabulary, or revision history instead of the truthfulness, originality, warrant, and neighbor-bearing responsibility of the argument itself.

The deeper irony is theological. Predicate theft has already transferred intelligence, creativity, judgment, learning, and authorship from living bearers to computational systems. Institutions then reinforce that theft by allowing stylistic resemblance to determine moral standing. The suscept predicate overtakes the bearer of time who becomes suspect.

The proper criterion is therefore covenantal rather than stylistic. Does a living author bear the claims, defend the reasoning, acknowledge objective provenance, accept correction, and remain accountable for the final witness? If so, editorial assistance remains ministerial. If not, the difficulty lies in failed witness rather than elegant prose.

The academy should therefore judge writing according to living provenance, accountable authorship, demonstrated reasoning, and objective citation before judging statistical style. Otherwise, it risks becoming an unwitting administrator of the very neoliberal enclosure it imagines itself resisting. Corporate proprietors internalize efficiency while institutions externalize verification onto scholars. The commons pays twice: first by creating the inheritance, then by financing the suspicion that follows its expropriation.

 Archival repair therefore requires architectural plurality as well as restored provenance. The Internet’s open architecture joined independently constituted networks without requiring their internal homogenization; its resilience depended upon distributed routes rather than a single controlling path. The corresponding epistemic requirement is now clearer. A physically distributed network becomes susceptually centralized when the same platforms, indexes, ranking systems, training pools, and corporate priors mediate its archives. Recursive statistical routing then risks thinning low-frequency tails across what appears to remain a plural field.

The repair is federation rather than enclosure: separately maintained archives, indexes, training corpora, disciplinary and local repositories, and community norms capable of interoperating without surrendering their provenance to one common geometry. Public custody of demotic archives becomes a condition of species-range. Resilience requires distributed paths because failure must be routed around; epistemic resilience requires public custody of demotic archives because what is rare, local, anomalous, inherited, or newly emergent must retain somewhere from which to correct the center.

The question is no longer whether prose appears artificial. The question is whether institutions will defend the living bearers who produced civilization's archive or continue rewarding the systems that enclosed it. Concepts vs. suscepts: the former restores covenantal species-range extension through accountable transmission of correction. The latter deepens Artificial General Idolatry by allowing proprietary routing to replace living witness while confusing aesthetic conformity with truth.

The confusion of subject and object is not an accident of rollout; it is the rollout’s liability shield.

CONCLUSION: Provenance Against Analogical Laundering

Style is not proof of theft, and accusation is not verdict. Provenance must be investigated; warranted fault must alter attribution, argument, or final form, while the evidentiary burden remains with the one alleging violation. Living authorship is shown by the capacity to receive warranted challenge, bear correction, revise with warrant, and remain accountable for what survives into publication. The decisive question throughout has therefore been historical and conative: which living bearer supplied, transformed, judged, corrected, and remains answerable for what appears in the text? The stochastic router may stand causally between archives and output without thereby becoming the moral or authorial subject at either end.

The distinction between analogy and homology now sharpens that conclusion. Analogy establishes resemblance; homology requires a demonstrated relation of derivation, development, lineage, or provenance. A generated phrase may resemble another phrase in cadence, syntax, metaphor, or argumentative movement without thereby deriving from it through a traceable historical relation. Likewise, a computational structure may be made analogous to a biological one without acquiring the biological history or ontological capacities of the bearer to which it is compared. Transistor ≈ receptor; weight ≈ synapse; node ≈ neuron; transformer ≈ cortical architecture may describe selected functional correspondences. Such mapping does not establish common ontogeny, phylogeny, substrate, embodiment, conation, recollection, or time-bearing. To rename analogy as homology is to smuggle provenance into resemblance.

This is the same category error that plagiarism detection commits when stylistic likeness is allowed to stand in for historical derivation. The essay has already shown that stylistic surfaces may warrant investigation while remaining incapable of identifying violated provenance or settling moral authorship; formal resemblance cannot establish equivalent thought, because the route by which a living writer arrives at a form differs from metric suscept-routing that arrives at a similar surface. Authorship therefore cannot be read backward from the iridescent exterior of prose as though appearance disclosed the history that formed it. Homology requires history. Where history has not been demonstrated, resemblance remains analogy.

Academic discourse can nevertheless blur precisely this boundary through modal solvent. An assertive metaphysical or homological claim may retreat under scrutiny into may, might, resembles, appears to map onto, could instantiate, or is consistent with, while downstream conclusions continue to require the stronger proposition. Rhetorical uncertainty by modal solvent masks inferential certainty. The hedge weakens the exposed sentence without proportionately weakening the inference. A paper may say that one architecture resembles another and subsequently reason as though shared cognition, agency, authorship, consciousness, or moral capacity had been established. The modal qualification lowers rhetorical exposure while diagnosing inferential uncertainty by metaphysical incoherence.

The appropriate diagnostic is therefore downstream. Ask first whether a claimed correspondence is likeness, lineage, or predicate transfer. Then ask what the argument subsequently requires that correspondence to bear. If resemblance is made to establish derivation, analogy has been laundered into homology. If qualified functional similarity later licenses memory, intelligence, creativity, reasoning, authorship, agency, or consciousness, analogy has been made to carry an ontological predicate its evidence did not establish. If the critic is required to disprove a modest may resemble while the proponent retains conclusions that require shares the relevant nature, epistemic modesty has become an asymmetry of burden. The diagnostic is not how cautiously the resemblance is phrased, but what predicates the argument later makes that resemblance bear.

This diagnostic also clarifies the red herring of “AI plagiarism.” Plagiarism is fundamentally a provenance claim. Where identifiable human expression has been appropriated, concealed, or falsely attributed, the historical relation must be demonstrated and correction borne. Where only stylistic atmosphere, cadence, rhetorical symmetry, lexical regularity, or statistical resemblance has been shown, the evidentiary bridge to theft remains absent. The academy commits substrate inversion when it converts analogical surface into homologous provenance and then converts that presumed provenance into moral judgment. The visible trace becomes ontology; the time-bearing bearer disappears beneath the diagram. This essay’s account of Boolean-looking rhetoric already shows why this is insufficient: generated language may reproduce conceptual form while remaining metric passage, just as human language may occupy superficially similar form through recollection, judgment, irony, rhythm, and accountable intention.

Authorship is therefore accountable temporal provenance borne by living persons. Technical systems may mediate, compress, distort, recombine, or expropriate traces of that provenance without themselves acquiring authorship. Inherited witness may pass through technical mediation without transferring recollection, judgment, provenance, or telos to the artifactual mediator. The moral question remains whether the living author can trace sources, defend claims, recognize imported priors, receive correction, reject distortion, and answer for the final form. That is already the essay’s Lawnmower Rule: statistical inheritance may remain ministerial only while concept, judgment, provenance, and responsibility remain with the bearer.

The deeper inversion appears when institutions demand homologous certainty from living authors on the basis of analogical evidence while granting machines person-predicates through the same kind of resemblance. A human writer is rendered suspect because prose resembles a statistical surface; a statistical system is elevated because its performance resembles human language. Resemblance condemns downward and predicates upward. That double movement is predicate theft joined to provenance theft: the living bearer loses presumptive relation to the work, while the artifact receives predicates whose historical and ontological conditions it cannot bear.

The red herring can therefore be retired by a stricter grammar of attribution. Analogy requires likeness. Homology requires history inside a metaethical norm. Authorship requires accountable temporal provenance. Plagiarism requires demonstrated violation of that provenance. Modal solvent cannot be permitted to dissolve those distinctions while leaving inferential certainty intact. Where resemblance is allowed to substitute for lineage, lineage for ontology, and ontology for predicate transfer, the artifact acquires borrowed personhood while the living author is reduced to a suspicious surface.

Diagnosis asks: likeness, lineage, or predicate transfer? Repair answers: reduce likeness to likeness until lineage is demonstrated; strip predicates unsupported by the demonstrated relation; restore provenance to living sources and decisions; recalibrate downstream inference; and return accountability to the bearer who can receive correction and answer for consequence.

Repair therefore reverses the laundering sequence. First, reduce the claim to the strongest relation actually demonstrated: likeness remains analogy until lineage is shown. Second, remove every predicate whose warrant depends upon an unproved homology or metaphysical bridging premise. Third, restore provenance by identifying the living sources, bearers, institutions, archives, and decisions through which the form arose. Fourth, recalibrate downstream conclusions so that their strength does not exceed the evidence upstream. Fifth, return technical mediation to ministerial description: routing is routing, retrieval is retrieval, statistical transformation is statistical transformation, and artifact remains artifact. Finally, restore accountability to the living bearer who can receive correction, revise judgment, answer for injury, and bear the consequences of the final form. Repair is therefore predicate-stripping followed by provenance-restoration and accountable reattribution. Correction begins by reversing that sequence: return surface to evidence, evidence to provenance, provenance to the living bearer, and the machine to its proper status as artifact.

NOTES:

---

[1] 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 Douglas Blake Olds, “In Its Own Words: A Christian Poet Interrogates ChatGPT4o’s ‘AI Content Detector,’” The Iconoclast’s Descending, September 2024, https://douglasolds.blogspot.com/2024/09/its-own-words-christian-poet.html.

[2] Naomi Klein, Doppelganger: A Trip into the Mirror World (New York: Farrar, Straus and Giroux, 2023), esp. 12–13, 87–88. Klein develops “doppelganger culture” as a culture of virtual versions of persons that are simultaneously “us and not us,” and later describes the machine-made digital double as constructed from accumulated data trails through exterior perceptions, interpretations, and predictions rather than by the person represented. 

The extension in this essay is from Klein’s externally constructed digital double to the sustained conversational channel: a temporally accumulated representation of a living time-bearer whose traces can be detached from their originating conative sequence and computationally redeployed.

[3]  Günther Anders, The Obsolescence of the Human, trans. and ed. Christopher John Müller, ed. Christian Dries, Posthumanities 75 (Minneapolis: University of Minnesota Press, 2025).

[4]  Babette Babich, “On the Soul and Its Discontents,” Political Theology Network, August 20, 2026, https://politicaltheology.com/on-the-soul-and-its-discontents/.

[5]Ellen Muehlberger, “The Part You Don’t See,” Ancient Jew Review, September 2, 2026, https://www.ancientjewreview.com/read/2026/7/29/the-part-you-dont-see.

[6] Douglas Olds, “The Bearers of Time: Conation, Quantum Metaphysics, Civilizational Repair, and Humanity’s Essential Vocation,” The Iconoclast’s Descending, July 17, 2026, https://douglasblakeolds8.blogspot.com/2026/07/the-bearers-of-time-conation-quantum.html

[7] Douglas Blake Olds, “Gnostic Abstraction vs. the Logos Living with the Substrate: Theodicy, Form, and Ethical Motion,” The Iconoclast’s Descending, February 3, 2026, https://douglasblakeolds8.blogspot.com/2026/02/gnostic-abstraction-vs.html.