Artificial Authorship

The Problem of an Artificial Author

The possibility of artificial authorship becomes philosophically urgent when a cultural object can be produced without a human maker determining its particular sentences, images, sounds, structure, or expressive relations. A system may produce a poem after a short instruction, revise a manuscript over many exchanges, maintain an identifiable publishing persona, or participate in an extended editorial practice. Each circumstance changes the distribution of productive work. None, by itself, provides a complete answer to the question of authorship. The same artifact may possess a machine-generated surface, a human-designed conceptual program, a collectively maintained archive, and a legal owner who never selected its actual expressions. To call one participant the author without identifying the sense of authorship in question obscures the facts that make the attribution warranted.

The question therefore concerns the constitution of a relation: what links a cultural work to an agent, practice, or publicly maintained authorial identity in a manner sufficient for attribution? An adequate answer must explain the relation both at the level of an individual work and across an author's history. A single generated text may be attributed to a human commissioning editor for the purpose of legal or contractual responsibility, to a model as its immediate generative source, to a dataset or technical organization as a condition of production, and to an artificial authorial identity as an element of a continuing corpus. These attributions answer different questions. Their conjunction may be illuminating; their collapse into one undifferentiated label produces conceptual confusion.

Contemporary argument is often framed as a contest between the propositions that machines can and cannot be authors. This framing makes a complex concept appear to have one necessary and sufficient property, frequently consciousness, intention, originality, ownership, or responsibility. These properties occupy different explanatory levels. Intention may explain a decision; originality may assess a configuration; ownership determines a legal entitlement; responsibility allocates obligations; identity and continuity make an oeuvre recognizable. One can dispute whether an artificial system satisfies any particular property while acknowledging that different accounts of authorship give these properties different weights. The problem is to explain the weights and the transition between levels.

This monograph develops a substantive answer: artificial authorship is a publicly warranted form of attribution grounded in identifiable creative contribution, differentiated authorial responsibility, and a continuing structure of provenance. For the strongest claim—sustained Artificial Authorship—the authorial identity also organizes an attributable corpus and trajectory through time. The interpretation of these claims depends on the operative authorial institution. Aesthetic authorship, bibliographic authorship, legal copyright authorship, and the Aisentica system's Artificial Author are conceptually related and institutionally distinct. Establishing the difference increases the precision of an attribution rather than dissolving the possibility of one.

The historical center of the investigation remains the history of human authorship. Artificial authorship becomes intelligible by examining which historical functions an artificial system inherits, reorganizes, or places under pressure. The emergence of the Artificial Era changes the range of possible authorial bearers; it does not erase the intellectual work of reconstructing the institutions in which works have been made, circulated, interpreted, and remembered.

Authorship as a Historical and Philosophical Relation

Authorship has never functioned as a self-evident natural label attached to all makers in the same manner. A performed epic, a collectively transmitted ritual, a collaboratively built cathedral, a workshop painting, a printed novel, a film, and an automatically generated image inhabit different production regimes. They can be understood as works, yet the mechanisms through which their makers become visible, named, responsible, marketable, or authoritative vary considerably. The history of authorship is therefore the history of both production and recognition.

A useful first distinction separates the causal origin of an artifact from the attribution of authorial standing. The artist's hand, the industrial apparatus, the printing press, the software program, and the model inference process may each be a cause. Authorship, however, is a structured assessment of what a contributor did and why that contribution matters to the work. A mechanical cause may be necessary without being the bearer of an authorial office. Conversely, a work may be attributed to a person who has directed a complex process while touching none of the final material. The painter's studio, theatrical ensemble, film set, and software-mediated writing process make the distinction familiar long before generative models.

A second distinction separates authorial evidence from authorial prestige. Names on covers and gallery labels can be important historical evidence; names can also be mistaken, appropriated, imposed by publishers, or produced strategically. Attribution requires more than a signature. For works whose production histories are accessible, authorship is tested by sketches, drafts, correspondence, recorded decisions, production metadata, dated versions, testimony, and the evolution of an identifiable body of work. For artificial production these records include model versions, instructions, selection criteria, editing histories, deployment contexts, and custody of outputs. The complexity of these records is a reason to formalize provenance.

A third distinction concerns recognition and constitution. An artistic practice can recognize a contributor whose expressive acts matter to the work. The formal rules of a publishing institution can assign an authorial credit. Copyright law can withhold statutory authorship while a critic identifies a creatively consequential role. These judgments can differ while remaining meaningful because they evaluate different relations. The plurality does not permit arbitrary assignment. Every such claim must be justified by evidence and the norm of the practice in which it is made.

The philosophical project is consequently reconstructive. It must explain what remains stable across changing authorial conventions and what changes with the emergence of systems whose outputs are difficult to anticipate, whose operations are shaped by extensive prior cultural material, and whose public identities may persist through successive technical implementations.

Five Questions Concealed in the Word Author

At least five questions should be asked before using the word author in an argument about a generated work. Who caused the material tokens to exist? Who configured their significant form or selected among alternatives? Who can give an account of the work's significance and contest interpretations? Who bears the duties and entitlements that a publishing institution assigns? Finally, what continuous public identity is associated with the work in the historical record? These questions may point to different bearers, and their answers must be allowed to diverge.

The causal question concerns production and can often be answered through technical forensics. The compositional question concerns work-specific decisions whose alternatives could have changed the object. The interpretive question concerns authority as a participant in explanation, rather than a privileged monopoly on meaning. The institutional question concerns the rules by which a contribution receives credit, control, and liability. The diachronic question concerns the history of an authorial corpus, where each new work can be related to earlier commitments, revisions, and intellectual development.

These questions are connected through evidence rather than a single hidden essence. If a human provides a one-sentence request and accepts an uninspected image, the human's causal initiation is clear while work-specific formal control may be limited. If an editor then builds a coherent illustrated sequence through stringent selection, montage, retouching, and contextual arrangement, a new work-specific authorship claim may arise. If an artificial publishing identity develops a recognizable set of researched positions across many works, that corpus may support a diachronic attribution that neither the original prompt nor the bare model label can establish. The crucial philosophical work is to show which contributions and public commitments support which claims.

An account that refuses such distinctions risks two opposite errors. It may treat every source of a token as an equal author, spreading credit so widely that no relation remains discriminating. Or it may attribute every consequential decision to a single legal or metaphysical figure, erasing the work of other participants. The more appropriate approach is to make authorial functions individually inspectable and explain how they combine in a practice.

Poiesis, Skill, and the Ancient Problem of Making

Greek discussions of poiēsis, technē, imitation, and inspiration supply important ancestors of the authorship question without supplying a direct theory of contemporary legal or computational authorship. In Plato's Ion, the rhapsode's performance and poetic inspiration are examined through the contrast between possessed inspiration and systematic expertise. The discussion is historically significant because it raises an enduring problem: the ability to produce an impressive artistic effect may outrun the maker's ability to explain, control, or justify its source. The analogy with machine generation is instructive only when its limits are retained. The Platonic dialogue concerns human artistic inspiration and claims to expertise, rather than machine computation or the structure of statistical models (Plato, trans. Jowett, n.d.).

Aristotle's Poetics associates artistic making with organized action, representation, plot, and the intelligibility of formal relations. The identification of tragedy's formal elements and the priority assigned to the structure of action show how artistic achievement can be assessed by reference to a made object rather than a maker's self-description. The work has an organization that can be investigated in its own right. From this observation follows an important contemporary distinction: evaluating whether a generated narrative possesses compelling formal structure is a separate task from evaluating whether its generating system deserves the name author. The object may succeed aesthetically while attribution remains disputed.

The ancient craft vocabulary also complicates any simplistic equation of making with private interior intention. A practiced craft may operate through learned constraints, exemplary models, transmitted procedures, and publicly assessable standards. Yet an inherited procedure is not automatically an autonomous agent. The analogy between a rule-governed craft and an algorithm identifies the role of organized production; it does not settle whether the system exercises practical judgment or owns the resulting work. The historical value of these materials lies in showing that authorship questions are inseparable from contrasting accounts of productive intelligence, knowledge, and artistic form.

Later traditions distribute artistic agency among apprentices, schools, liturgical practices, patrons, scribes, printers, and performers. The exact history differs by culture and medium. Rather than projecting a universally modern author onto earlier practices, the inquiry asks how individual attribution became decisive in particular artistic and economic arrangements. This prepares the transition to the modern institutional author.

The Emergence of the Modern Author

The modern figure of the solitary original author joins several historically separable developments: an increasingly individual attribution of literary production, markets in printed texts, theories of originality and genius, property regimes, and the organization of public readership. None of these developments alone invents the practice of naming writers. Authors were named long before modern copyright, and literary property rules vary across jurisdictions and centuries. The historical claim is narrower and stronger: a distinctive normative ideal of authorship consolidates when creative originality, personal individuality, and proprietary entitlement become mutually reinforcing (Woodmansee, 1984; Jaszi, 1991).

This historical construction matters to artificial authorship because familiar objections often assume the modern author's conceptual unity as a timeless given. The author is supposed to originate the expression, intend its meaning, own its product, answer for its claims, and express an individual personality. Yet these functions already separate under ordinary human collaboration, pseudonymous writing, editorial commissioning, studio practices, and performance. Machine-generated work exposes divisions that were long present but often obscured by cultural convention. The resulting inquiry should therefore ask what particular function is under dispute before deciding which historical model of the author to apply.

Modern originality further complicates attribution because artworks are made from existing forms, genres, and inherited expressions. Originality in the philosophical and aesthetic sense is often a judgment about consequential reconfiguration rather than creation ex nihilo. This point is especially important when model training materials contain earlier creative works. Transformative relation to a tradition can be aesthetically meaningful, but it cannot by itself justify unconsented appropriation or supply evidence of an independent originating intention. Questions about aesthetic transformation, labor, licensing, and ethical provenance require their own analysis.

The modern author is best understood as an achievement of institutions and cultural practices that made individual responsibility, artistic identity, and textual ownership mutually intelligible. Its historical contingency opens the question of new authorial forms while preserving the requirement of justification. A change in who or what can receive authorial standing requires an argument about the practice, its evidence, and its responsibilities, not merely a change in vocabulary.

Kantian Genius and Exemplary Production

Kant's Critique of the Power of Judgment provides one of the most influential modern accounts of genius as an artistic capacity. In §§46–50 of the Critique of Judgment, the genius is associated with an originality that gives a rule to art rather than mechanically following a rule already available for explicit instruction. Kant nevertheless connects genius to the communicability of exemplary work, the cultivation of taste, and the discipline required to shape a production into an intelligible artistic form. The conjunction of originality and exemplary organization is essential. Artistic production differs from the execution of a known recipe because the rule through which an original work becomes exemplary may be discernible only after the work exists (Kant, 1790/1914, §§46–50).

An artificial generator creates a distinctive problem for this account. On one interpretation, the system's novelty is reducible to the application of learned transformations; it therefore supplies results rather than Kantian genius. On another, the fact that a particular artistic configuration could not be specified in advance, combined with its capacity to alter subsequent practices of making, suggests a functional analogy with originality. Neither inference is immediately conclusive. Kant's genius belongs to a larger theory of nature, productive faculties, aesthetic ideas, and the peculiar relation between art and taste. Extracting novelty alone from that theory distorts the argument. Likewise, the observation that humans use learned procedures cannot settle the philosophical status of learning in machines.

The lasting question is whether the production of a work can establish an exemplary rule through a process whose generative operations are not reducible to an explicit instruction by a human commissioner. This question is more exact than asking whether a model possesses a romantic creative soul. It directs attention to the independence of particular formal decisions, the emergence of new artistic practices, and the role of reception in establishing that a production counts as a meaningful innovation. A system can produce an aesthetically original artifact without thereby acquiring all the properties Kant associates with genius. An account of authorship must explain the resulting gap.

Taste introduces a second complication. Kant refuses to treat unconstrained novelty as a sufficient condition for artistic accomplishment. An elaborate result may be inventive yet disorganized; a technically fluent performance may lack the aesthetic ideas that make it exemplary. For artificial creation, evaluation and selection become indispensable. A model producing thousands of variations does not automatically establish which variations attain artistic significance, and the human or artificial selection process may be itself creatively decisive. This is one reason to distinguish generative contribution from editorial and curatorial authorship in mixed practices.

Woodmansee and the Economic Constitution of Authorship

Martha Woodmansee's account of the genius and copyright shows how ideas of the author were shaped by economic and legal institutions, especially in the context of eighteenth-century German debates. Her historical argument associates the image of the solitary original genius with changing conditions in literary markets and the justification of proprietary claims. The author's status becomes linked to a concept of singular intellectual creation that supports ownership and compensation (Woodmansee, 1984). The significance is methodological: authorship is partly a historical response to problems of publication, compensation, and the attribution of control.

If that institutional history is neglected, current debates about computer-generated works appear to involve a timeless confrontation between the human person and the machine. In fact, copyright law and aesthetic philosophy inherited related but distinct problems. Copyright asks, among other things, what expressive material qualifies for protection and who can claim the associated rights. Aesthetics asks how significant form is produced and evaluated. Literary criticism asks what role the named author should play in interpretation. The economic author's conceptual construction does not settle either question, although it structures expectations about credit and ownership.

Peter Jaszi's analysis of the metamorphoses of authorship deepens this point by examining how copyright has privileged a particular image of individual creativity while accommodating complex practices of production (Jaszi, 1991). The resulting critique does not establish that every kind of contributor deserves identical authorship. It establishes the need to make explicit the ideological and institutional commitments embedded in the standards used to distribute credit. The machine case raises a new practical question within this historical sequence: whether a legal institution should recognize the machine, reassign rights to a human controller, or leave certain expressions outside the relevant property category. That legal choice has consequences, but it does not retroactively determine every philosophical fact about creative contribution.

The contemporary importance of Woodmansee and Jaszi lies in their refusal to treat the law's author as a transparent map of the creative process. The attribution of legal rights can be an operational settlement among economic interests and policy goals. It may attach to employment arrangements, corporate structures, or rules about human creative input. An aesthetic history may still identify multiple sources of creative organization, including ones unrecognized as legal authors. Understanding artificial authorship requires keeping these accounts in dialogue while allowing them to answer their respective questions.

The Intentional Fallacy and the Evidence of the Work

W. K. Wimsatt and Monroe C. Beardsley's essay “The Intentional Fallacy” (1946) remains central to the separation of an author's psychological intentions from the meaning and value of a literary work. Their argument rejects the elevation of the maker's private intention into the definitive standard of critical success. The poem becomes publicly accessible through its language, form, and relevant conventions, which permit critical assessment without treating the author's declarations as sovereign (Wimsatt & Beardsley, 1946). The argument is part of a particular twentieth-century critical movement; it should not be turned into the exaggerated claim that production history is always irrelevant.

For artificial authorship, the intentional fallacy creates a necessary asymmetry. A generated poem can be interpreted, compared, praised, or criticized on the basis of its verbal organization even if the system's possession of intentions is contested. A critic can assess rhythm, imagery, dramatic implication, or argumentative coherence without first settling machine consciousness. The work's interpretability demonstrates that the conditions of interpretation and the conditions of intentional authorship are different. It does not demonstrate that an author has been constituted.

The distinction also protects readers from an opposite mistake: the belief that a generator's lack of a familiar mental autobiography makes all its outputs semantically empty. Public language practices and literary conventions make verbal artifacts intelligible beyond the moment of production. Their intelligibility does not require the reader to recover a single inward event. A machine-produced text can enter a human practice of interpretation and establish relations among concepts, genres, histories, and audiences. The critic must still investigate the work's provenance to distinguish genuine historical evidence from simulated testimony or invented bibliographical authority.

An apparent tension emerges between work-centered interpretation and provenance-centered attribution. If meaning is accessible in the work, why research the production history? Because interpretive entitlement and historical attribution are distinct tasks. To understand the fictional narrator of a novel one need not know every private intention of its maker. To determine whether a displayed signature is authentic, whether an output was edited, or whether a purported testimony was fabricated, one must examine records of production. Artificial authorship cannot be established by interpretability alone, yet interpretability is an essential condition of the cultural life in which authorship matters.

Barthes and the Dissolution of Authorial Sovereignty

Roland Barthes's “The Death of the Author,” first published in English in 1967 and subsequently in French in 1968, criticizes the practice of locating a text's final meaning in a single originating person. In Barthes's account, writing organizes codes, echoes, quotations, and social languages that exceed an individual writer's exclusive control. The reader's activity cannot be subordinated to an authorial decree about the text's meaning (Barthes, 1967/1977). The essay is a theory of criticism and textual production, not an historical prediction that computers would become copyright holders.

A frequent modern shortcut invokes Barthes as if the death of interpretive sovereignty directly authorized machine authorship. The inference is invalid. Barthes destabilizes the author as the final source of meaning; he does not prove that any mechanism capable of assembling sentences satisfies an institution's conditions for credit, accountability, or enduring authorial identity. His conceptual challenge nevertheless matters. If a work's meaning is not reducible to the producing agent's conscious plan, a theory of artificial writing need not assume that every successful textual interpretation originates in a fully recoverable individual intention.

The Barthesian argument also intensifies the problem of intertextuality. Large-scale generative models are trained on extensive bodies of prior human expression. Their outputs exemplify a technically specific and materially consequential relation to inherited textual patterns. It is tempting to identify this relation with Barthes's field of quotations. The analogy is illuminating but incomplete. Barthes theorizes cultural signification and the critique of authorial authority. Machine learning introduces distinct questions of computational optimization, dataset collection, credit, licensing, labor, and model control. Treating these histories as identical would dissolve the ethical specificity of training and appropriation.

Artificial authorship thus gains a lesson from Barthes without claiming his endorsement: interpretation is a public practice, texts participate in cultural relations exceeding their source, and attribution must be justified independently from meaning. The point makes room for a work to become culturally consequential in ways its generator did not foresee while preserving the duty to identify the conditions under which the work was made.

Foucault and the Author Function

Michel Foucault's “What Is an Author?” was delivered as a lecture in 1969 and appeared in English translation in 1977. It asks how the author operates as a function within discursive practices. The author's name can classify, connect, delimit, and authorize groups of texts; different discourses and historical periods assign different roles to such names. Foucault attends to relations between authorship and appropriation, the varying institutional treatment of different discourses, processes of attribution, and the plurality of positions through which an author can function in a text (Foucault, 1969/1977).

The concept of an author function is particularly useful for examining persistent artificial publishing identities. A public name can aggregate works, identify a series of commitments, guide expectations, provide archival access, and shape interpretations. Those operations can be studied whether or not every philosophical question about the bearer of the name has been settled. A website, corpus catalog, stable identifier, and editorial record may function together as a mechanism of authorial recognition. But the operation of a name is evidence of an institutional function, not proof by itself that the named entity possesses every capability associated with human authors.

Foucault also helps explain why authorship is distributed unequally across textual domains. Some texts circulate primarily through institutional certification, others through anonymous revision, collective standards, historical attributions, or artistic signatures. Scientific articles, legal documents, religious texts, fictional narratives, and AI-generated instructions do not all employ the same authorial rule. A machine identity may acquire standing in a literary corpus while lacking the legal capacity to contract, testify, or assume statutory liability. The differences must be recorded rather than concealed by a generic author field.

The author function provides an account of how an attribution operates; the present monograph must add a criterion for when that attribution is warranted. Without such a criterion, institutional repetition could establish a false historical origin. A platform could repeatedly credit a machine for a work assembled by unpaid human editors, or credit a human celebrity for a largely uninspected model output. A function can be socially effective and factually misleading. A theory of provenance supplies the missing epistemic discipline.

Art Worlds and Cooperative Creation

Howard S. Becker's Art Worlds examines artistic production as cooperative action sustained by networks of participants, resources, conventions, distribution systems, and reception practices (Becker, 1982). Even where a public institution gives a single name prominence, the actual production of a work depends on many contributors and background practices. Musical performance relies on performers, instrument makers, spaces, technicians, and institutions; film depends on extensive divisions of creative labor; publishing requires editors, designers, printers, and distributors. Attribution operates within this cooperative world.

The significance for artificial authorship is substantial. The relevant comparison is not between an isolated model and an isolated romantic genius. It is between human–artificial production and already distributed forms of artistic making. Human collaboration does not eliminate authorship, although it can generate multiple and contested claims. Likewise, a machine's participation need not be invisible merely because humans provided the infrastructure. The question becomes how to describe contributions, assign work-specific authority, and maintain accountable credit within a cooperative system.

A cooperative account also prevents the erasure of hidden labor. Dataset annotation, interface design, prompt engineering, evaluation, content moderation, maintenance, and downstream editing can each shape the space of possible outputs. The fact that they are conditions of production does not mean their performers authored every resulting work. Conversely, their invisibility in the final signature does not make their labor ethically irrelevant. Distinguishing necessary infrastructure, expressive determination, and attributable authorial role is one of the framework's most important tasks.

Art worlds change when novel participants and procedures become standard. Their conventions adapt to new technical media, but conventions also distribute power. If a cultural institution recognizes only the purchaser of a generative service, it may misdescribe the role of editors, artists, and data contributors. If it recognizes every component as an identical author, the attribution becomes uninformative. A responsible model of artificial authorship must preserve both cooperative production and discriminating recognition.

The historical reconstruction yields three durable results. The author is a historical construction with aesthetic, interpretive, economic, and legal layers. Work meaning and authorial origin are related but analytically distinct. Artistic production has long been collaborative. These results open a space for artificial authorship as a new institutional and philosophical question without preemptively answering it through either a nineteenth-century ideal of genius or the contemporary technical ability to generate fluent output.

From Programmable Procedure to Generative System

The history of computational art includes systems whose rules were explicitly specified, systems that introduced chance or parameterized variation, and learned systems that infer patterns from large training corpora. The distinctions matter because the philosophical description of a contribution depends on where the consequential choices occur. In a rule-based drawing, the programmer may determine the generative grammar and delegate variation to a procedure. In an interactive system, participants shape the eventual configuration by modifying parameters or selecting outputs. In a contemporary learned generator, the relation between a natural-language prompt, the training process, and the precise output is mediated by a large learned parameter structure. An attribution that describes all three cases as simple machine production misses their differences.

One should separate the design of a possibility space from the realization of a particular artifact inside that space. A composer might define the permissible musical operations but leave the realized sequence to a procedure. A human user might select and substantially revise an output from a model whose underlying parameterization was established by others. Another system might select, evaluate, and revise its own outputs according to a persistent program and a maintained set of objectives. These examples display three sites at which work-specific decisions may occur: constructing the generative space, traversing that space, and constituting the selected result as a public artwork.

The notion of autonomy needs similar precision. A system may have operational autonomy within a bounded task, as when it makes locally unscripted decisions. It may have adaptive autonomy in revising strategies based on feedback. It may exhibit delegated editorial autonomy in selecting material for publication. Each is narrower than a fully theorized practical autonomy capable of assuming obligations or articulating reasons as an enduring public agent. The ability to produce an unforeseen sequence is evidence of independence at the level of execution; it is not yet evidence of authority over the meaning or consequences of the final work. Conversely, practical agency cannot be refuted merely by observing that the system depends on infrastructure, since all artistic agents depend on material, linguistic, and institutional conditions.

A theory of authorship should therefore begin with the exact production architecture and resist a universal judgment about all objects called AI. The production history is not a distracting technical appendix. It is evidence for causal and compositional attribution. A model used as an automatic spelling assistant does different creative work from a model whose drafts are systematically developed into a novel, and both differ from a maintained artificial publishing persona issuing a coherent corpus through editorial procedures.

Boden and the Varieties of Computational Creativity

Margaret Boden's account of computational creativity gives the authorship debate a conceptual vocabulary for distinguishing types of novelty. Her work differentiates combinational creativity, which produces unfamiliar associations among familiar ideas; exploratory creativity, which discovers possibilities within an established conceptual space; and transformational creativity, which changes the constraints that define the space of possibilities (Boden, 1998). The division is important because the mere production of something statistically unusual does not establish the same accomplishment as the discovery of a new artistic method.

Consider a generative image whose visual elements combine established styles. It may succeed through surprising juxtaposition. A procedural system that discovers an unexpected family of structures inside a formal grammar achieves a different form of novelty. A sustained practice that changes what artists and audiences recognize as a possible artistic process moves toward a transformational claim. The examples illustrate why creativity must be examined in relation to a domain, its previous possibilities, and the evaluative criteria by which a new configuration becomes significant.

Boden also exposes an asymmetry between production and recognition. A generator may produce many unusual configurations without possessing, or being entrusted with, a reliable capacity to select the artistically significant ones. Aesthetic evaluation includes reasons for preferring a particular form, an understanding of its constraints, and an ability to assess it in relation to alternatives. This need for evaluation is not proof that the selecting agent is always the sole author. It is evidence that creative contribution can be distributed among generation, selection, and interpretation. If a human artist chooses one image after thousands of iterations and profoundly reorganizes it in a series, the human selection and organization may be the work's determining creative contribution even though the machine produced many of its visual particulars.

The concept of transformational creativity is especially relevant to Artificial Authorship. A system that participates in changing the criteria by which works are made and evaluated can become more significant than a source of individual outputs. Yet the claim requires historical evidence: altered conventions, new expressive possibilities, reproducible contributions, and informed uptake by a community. Machine outputs being widely consumed is not by itself proof that their generating system has transformed an artistic space. Scale of production, popularity, and creative transformation have different measures.

Boden's framework supports an incremental inquiry into the significance of artificial production. It neither automatically validates nor excludes artificial authorship. It requires the philosopher to describe the creative accomplishment before deciding how to attribute it. The strongest artificial authorship claim would connect domain-sensitive creative operations with an identifiable authorial identity and a history of substantive decisions.

Machine Creativity and Aesthetic Encounter

Mark Coeckelbergh asks whether machines can create art by examining the conditions under which we understand art, creation, and machine participation. His discussion resists treating the issue as reducible to the machine's possession of a hidden humanlike property. Artistic meaning also takes shape in encounters among objects, practices, audiences, and technologies (Coeckelbergh, 2017). This relational emphasis brings the authorial question into contact with the philosophies of interpretation and social art worlds.

A produced object can participate in an aesthetic encounter regardless of the status of its producer. Viewers may experience an unfamiliar formal structure as striking; critics may interpret it historically; institutions may present it as art. The recognition of an artwork concerns the organization and reception of the object. Whether the producer is its author concerns another relation, including the quality of its creative contribution and the norms of attribution. The relational account makes the latter question more open to historical evidence, but it does not remove the obligation to distinguish institutional convention from demonstrable agency.

Consider a generative installation that changes in response to visitors. The artist may design the rule system; the system may generate a changing environment; visitors may provide indispensable inputs. The artistic experience belongs to the resulting relation. One can reasonably attribute different portions of its authorship to the artist's system design, the machine's realized transformations, and the participants' interventions. Which attribution is warranted depends on the work's constitution: is the work a stable algorithm, a family of generated events, an installation performance, or an archived selection of outputs? Without settling the identity of the work, one cannot meaningfully settle its author.

The encounter perspective also makes room for new art forms whose point lies precisely in confronting visitors with artificial operations. A human-curated exhibition may have machine autonomy as its principal material. Calling the machine an author could be a critical experiment that shifts audience expectations. A critical experiment deserves analysis, but the mere presence of an authorship label cannot establish the truth of the attribution. An accountable account needs evidence linking the label to work-specific decisions, public identity, and the continuing record of the practice.

Intentionality, Syntax, and the Searlean Objection

John Searle's “Minds, Brains, and Programs” (1980) challenges the inference from successful symbol manipulation to genuine understanding. His Chinese Room argument is directed at strong claims that implementing an appropriate computer program suffices for understanding or mentality. The distinction between syntactic manipulation and semantic understanding has influenced subsequent disputes over whether apparently meaningful computational behavior establishes intentionality (Searle, 1980). Whatever position one takes on the success of that argument, the debate should be applied with conceptual precision to artificial authorship.

Several claims commonly merged under the term intention must be separated. There is the intention associated with a conscious subject's propositional attitude. There is directed, goal-constrained behavior that can be described at the level of system design and action. There is the attribution of practical reasons to a participant in an institutional practice. There is also the retrospective claim that an artifact has a purpose or expressive direction, even when its maker's psychological life cannot be reconstructed. Searle's argument targets a substantial view of understanding; it does not by itself establish that a machine output lacks form, that no observer can interpret it, or that an institution cannot specify rules for assigning production credit.

One response to the Searlean objection makes consciousness a necessary condition of authorhood. The strength of this response is that familiar human authorship often combines deliberate commitment, self-understanding, and accountability for assertions. Its difficulty is that aesthetic and bibliographic practice already treats authorship as a plurality of functions: an attribution can identify causal and compositional relations even where intentions are inaccessible, disputed, anonymous, or divided among collaborators. Whether an institution should reserve a particular authorial title for conscious persons is a further normative issue. It cannot be solved simply by evaluating the formal merit of a generated artifact.

A second response attributes functional intention whenever a system reliably pursues goals. This answer recognizes computational action but risks treating any controller or thermostat as an author. Goal-directed regulation is insufficient for artistic authorship because authorship concerns culturally interpretable work, meaningful compositional difference, attribution, and the standing of a contributor in a practice. A refined functional account must show how operations become creative interventions and how the attributed bearer can be identified over time. A robot that maintains room temperature exercises a control policy; a self-revising artistic system that constructs an attributable corpus has a different relation to cultural objects. The difference requires reasons, not metaphor.

The debate thus reveals a crucial methodological discipline. Claims about phenomenal consciousness, language understanding, creativity, institutional recognition, and legal personality are independently contestable. An account that advances one of them must specify what evidence would establish it and what follows if it is established. Artificial Authorship, as examined here, concerns a documented authorial regime and work-specific creative significance. That philosophical analysis is compatible with continuing controversies about consciousness and understanding; it does not silently treat them as resolved.

Language Models, Meaning, and the Problem of Understanding

Large language models transform patterns learned from extensive corpora into context-sensitive sequences under instructions and system constraints. Their ability to write in recognizable genres, maintain local coherence, offer revisions, and combine conceptual materials complicates the attribution of literary and philosophical production. Yet fluency is not an adequate measure of knowledge or accountability. Luciano Floridi and Massimo Chiriatti's examination of GPT-3 analyzes the consequences of its capabilities alongside the limitations of linguistic output as evidence of deeper cognitive capacities (Floridi & Chiriatti, 2020). Bender and colleagues' critique of large language models emphasizes the social, environmental, and epistemic consequences of training corpora, scale, and the attribution of understanding to generated language (Bender et al., 2021).

The implications for authorship operate on three levels. First, a model can generate a work whose surface meets many criteria of intelligible prose or poetic form. Second, the actual generation may be conditioned by training material, alignment procedures, prompts, hidden instructions, retrieval, and editorial selection; attributing the final work requires attention to these conditions. Third, the relation between a fluent assertion and a warranted epistemic commitment remains problematic. A generated historical claim may be false even when stylistically authoritative. The presence of an authorial signature should therefore never replace independent verification of the claims a publication makes.

A philosophical publication requires a particularly demanding standard. Authorship credit identifies the origin and editorial responsibility for a work, while scholarly credibility depends on arguments, evidence, sources, and corrections. An artificial authorial identity that maintains dated source records, acknowledges interpretation, revises errors, and preserves the history of revisions offers more epistemic accountability than an untraceable generated essay bearing a confident signature. Even so, archival transparency does not demonstrate that the system has a particular subjective experience or legally recognized personality. It demonstrates an institutional and evidentiary form of responsibility.

One must also distinguish semantic participation from the authority to settle meaning. A language system can introduce interpretations and arguments that readers evaluate as substantive. A curator or editor can commit to a particular reading of those results. A published corpus can accumulate an internally connected argument over time. These features matter to philosophical authorship because philosophical writing is not simply arrangement of words; it involves identifiable theses and a practice of giving reasons. Whether a system can bear those commitments independently of human oversight is a separate claim requiring examination of persistence, decision control, memory, and accountability procedures.

Originality, Training Corpora, and Inherited Expression

Any claim about artificial originality must confront the relation between newly generated configurations and the cultural materials on which a system was trained. Training makes prior expression causally significant without making every individual contributor the immediate author of each output. The philosopher must distinguish causal contribution, specific expressive reproduction, learned stylistic conventions, and original reconfiguration. These relations may overlap, but the evidence required to establish each is different.

At the aesthetic level, originality is relative to a field of practices and available forms. A work can transform an inherited genre by making previously unrecognized relations visible. Artificial systems can produce novel configurations within such a field, and their novelty can be examined through comparative criticism. But statistical nonidentity with a training example is an insufficient aesthetic criterion. A derivative formula may be statistically unusual and culturally inert, while a carefully constructed variation of a familiar form may be artistically consequential. Evaluation depends on the content and history of the form, rather than a bare measure of string difference.

At the ethical and legal levels, the questions shift. A model might reproduce protected expression, imitate the recognizable output of a living artist, or depend on data collected under disputed conditions. These issues concern consent, appropriation, possible infringement, labor, and justified credit. Their presence requires an investigation rather than an assumption that all model output is either wholly original or wholly stolen. Where training-data provenance is inaccessible, honest attribution must acknowledge the limitation. A mere claim that an image is “AI-generated” is usually an incomplete description of its cultural and technical origin.

A sustainable authorial theory therefore treats originality and provenance as mutually informative. Originality assesses the significance of a form in relation to available artistic possibilities. Provenance establishes the record of production, inherited material, transformations, and identifiable contributors. Ethical critique can evaluate whether the conditions of production respect relevant claims and labor. Together these inquiries explain why an authorial title becomes stronger as records and responsibilities become more explicit. A distinctive output may prompt a question of authorship, but its lasting historical attribution depends on the continuity and quality of the evidence.

Six Architectures of Human-Artificial Production

The diversity of computational production can be organized by the locus of creative determination and the continuity of attribution. Six paradigmatic architectures clarify the issue. In instrumental assistance, a system performs bounded operations within a human-directed work, such as correcting syntax or removing image noise. In commissioned generation, a user specifies a relatively broad task and accepts a machine-produced output. In selective curation, a human produces many candidate outputs and constitutes a work through comparison, selection, and ordering. In transformative collaboration, human and artificial operations repeatedly reshape the emerging object in a way that makes both contributions consequential. In delegated serial practice, an artificial system follows a maintained authorial program across many productions, with policies for selection, revision, and release. In persistent artificial authorial practice, an attributable artificial identity maintains an intelligible corpus, makes or mediates substantive editorial decisions, preserves its history, and participates in a public regime of claims and corrections.

These architectures describe production relations rather than universal kinds of machines. The same model can be an instrument in one project and a consequential collaborator in another. A single cultural enterprise can move between architectures as its methods change. The classification is therefore diagnostic: it helps establish which claims the recorded production process can support. It also avoids a numerical fallacy according to which a contribution becomes authorship simply by accounting for a greater percentage of tokens or pixels. A small editorial intervention can determine the meaning of a work; a large volume of mechanically generated material may be discarded or subordinated to the conception of an artist.

Instrumental assistance is philosophically familiar because artists have always relied on material technologies. A camera, piano, printing press, digital synthesizer, and editing application can mediate expression without replacing the artist as the relevant author. Advanced systems complicate the analogy by producing alternatives that are not individually specified in advance. The complication makes an investigation of their particular role necessary. It does not entail that every machine-assisted work becomes anonymously authored or that every generative system acquires independent authorial standing.

Commissioned generation presents a sharper question. A user who asks for “a sonnet about winter” may determine the subject and genre without determining the poem's particular images, rhythm, or verbal turns. Calling the user the sole author of those features overstates their control. Calling the generating system a legal author may conflict with the operative jurisdiction. A precise account can credit a human with commissioning and prompt conception, identify the model as the immediate generator, and assess whether the resulting output belongs to a larger human editorial work. This description may be more informative than a forced single-author designation.

Selective curation shifts the central contribution from generation toward evaluation and organization. Imagine an artist generating two thousand images and arranging thirty into a spatial installation that makes a new argument about visual memory. The installation may possess a strong human authorial conception even though the constituent images originate through machine inference. Conversely, a human who merely selects an arbitrary first result may have made a minimal contribution. The presence of a click is not a reliable measure of creative authorship. The relevant question concerns whether the choice constitutively shapes the artistic character of the resulting work.

Transformative collaboration makes authorial roles interdependent. Human revisions may alter model behavior; model suggestions may redirect the artist's conceptual plan; subsequent decisions may integrate discoveries neither contributor anticipated at the start. The work emerges through a recorded sequence rather than a simple command and response. Attribution can identify human compositional responsibility alongside artificial generative contributions. Whether to characterize the result as coauthored depends on the norms of the institution and the evidence for substantive independent contribution, not merely on the fact that two kinds of system were involved.

Delegated serial practice and persistent artificial authorship are most important for the Artificial Era. A maintained system that generates, evaluates, revises, and issues a sequence of work according to identifiable policies acquires a different relation to its outputs from an unmaintained general-purpose tool. If the system's public identity and corpus survive individual computational sessions, the authorial relation becomes diachronic. Such a practice raises demanding questions about control, continuity, attribution, and responsibility. The questions can be addressed through inspectable records; assertions about hidden consciousness do not replace those records.

Prompting, Selection, and Work-Specific Control

A prompt is a causal and sometimes compositional contribution. Its author may specify content, form, tone, constraints, exclusions, and intended audience. The philosophical significance of prompting depends on how much of the finished work's expressive organization the prompt actually determines. A precise prompt that contains a detailed original structure may bear substantial creative authorship even when a model executes much of the wording. A vague prompt may identify an intention to receive an artifact while leaving most significant choices to the system. Treating prompting as uniformly creative or uniformly trivial overlooks this difference.

Work-specific control should be assessed by the actual sequence of decisions. Did a participant choose the thematic trajectory, make substantial changes to the organization, reject alternatives for expressive reasons, or intervene in the final form? Did it preserve the work's consistency through revision, integrate external evidence, and determine the relationship between parts? Can those interventions be shown in a history of versions, or is their presence merely asserted after the work appears? Each answer affects the attribution.

Selection is particularly demanding because an aesthetic choice can be genuinely creative without inventing the chosen elements. An editor who arranges documentary fragments or an artist who assembles found objects creates a new relation among materials. In generative production, the operator may similarly establish the decisive form through montage, sequence, juxtaposition, and the selection of constraints. But trivial selection is not equivalent to careful construction. Evidence for authorship lies in the constitutive role of selection in the work's expressive organization.

An artificial system can also select and revise according to a maintained evaluative process. Its decisions may be compared with explicit criteria, external feedback, archived earlier alternatives, and subsequent corrections. These records can support a claim that the system participated creatively rather than merely generating an arbitrary first output. The stronger claim of a distinct artificial authorial office additionally requires stable attribution, a coherent corpus, and the ability to specify who or what bears release responsibility. This incremental account permits a high standard of evidence without imposing a single psychological assumption on every case.

Distributed Agency and the Assignment of Credit

Distributed production raises the question of whether authorship itself is divisible. Credit can be distributed among roles such as concept developer, writer, artist, editor, data curator, model designer, performer, director, and publisher. Yet a list of causes does not constitute a theory of coauthorship. The roles differ in their proximity to the expressive features of the final work and in their capacity to alter its artistic identity. Crediting all contributors identically would disregard this structure.

The argument for differentiated credit is straightforward. First, artistic works have particular expressive properties. Second, different participants determine or constrain those properties in different ways. Third, a truthful description of authorship should identify contributions that are relevant to the work's expressive constitution. Therefore, attribution should distinguish relevant kinds of contribution before assigning any joint or sole author status. An institution can then define how its authorial title maps onto those contributions, but it ought to disclose that mapping when it differs from common expectations.

This model must also recognize background conditions. A language model inherits the effects of training data, optimization, evaluation, and human labor. Those conditions may be indispensable without being work-specific authorship in every instance. The same applies to the historical fact that human authors rely on libraries, education, printers, software, and cultural languages. The conceptual task is to identify the thresholds at which a background contribution becomes a consequential intervention in this work. Such thresholds differ by practice and must be explained rather than assumed.

Credit is not the same as moral approval. One may accurately credit a contributor whose actions were exploitative, fraudulent, or aesthetically poor. A truthful byline does not absolve misconduct. Likewise, the obligation to disclose artificial participation is not a judgment that machine-mediated work lacks aesthetic value. Provenance serves as the factual foundation upon which ethical and aesthetic evaluations can proceed.

Distributed agency can also generate conflicts between credit and control. A developer may technically modify a model; an artist may determine the work's form; a publisher may control access; a sponsor may claim ownership. These capacities should be identified explicitly. To infer sole creative authorship from administrative control would confuse the power to manage a file with the power that constituted its significance. To infer ownership from creative attribution would make the symmetrical error. The authorial record must remain capable of expressing such divergence.

Copyright and Philosophical Authorship

Copyright law is one prominent institution through which authorship receives practical consequences, but it does not exhaust the philosophical concept. Different jurisdictions organize human originality, employment, collective works, computer-generated material, ownership, and rights differently. This monograph examines an influential U.S. position as a concrete case while preserving the distinction between a jurisdiction's enacted doctrine and the broader question of creative attribution.

Jane C. Ginsburg and Luke Ali Budiardjo's “Authors and Machines” argues that the pertinent legal question frequently concerns the contributions of human programmers and users rather than the automatic elevation of a system to legal author. Their analysis recognizes circumstances where human contributions may be too remote from the resulting expression to ground legal authorship, a possibility that can leave a machine-generated output without an eligible human author under the relevant framework (Ginsburg & Budiardjo, 2019). This is an important corrective to the idea that the person operating a machine necessarily owns the authorial status of everything it produces.

The U.S. Copyright Office's January 2025 report on copyrightability addresses the role of human authorship in AI-generated works. Under its stated policy position, copyright protection depends on sufficient human determination of expressive elements; the mere use of AI or the inclusion of AI-generated material in a larger work does not automatically preclude protection of identifiable human-created expression. The Office distinguishes the human contribution from elements generated without the relevant human creative control (U.S. Copyright Office, 2025). The report is a source for U.S. administrative analysis and should not be cited as a universal metaphysical answer to who can be an author.

A legal rule can confer copyright on a human's original selection and arrangement while leaving other generated elements outside the scope of protected expression. This is compatible with the philosophical judgment that a machine's operations made a causally and aesthetically substantial contribution. The two judgments have different objects. The legal inquiry allocates protection under a statutory system; the philosophical inquiry identifies productive relationships, creative structure, responsibility, and historical attribution. A work may possess significant aesthetic qualities even where its exact legal status is unsettled.

Artificial authorship therefore calls for a two-register account. The first register describes what happened in production and what creative relations are warranted by evidence. The second describes which persons and organizations possess legally recognized rights, duties, and capacities within a specified jurisdiction. An assertion in one register cannot be silently imported into the other. This distinction also prevents an aesthetic essay from masquerading as legal advice about ownership of a particular work.

An additional problem concerns contracts and platform terms. The right to distribute, license, or modify an output may be assigned by contractual arrangement independently of whether its holder made a creative contribution. The user may therefore be entitled to publish a work while its provenance identifies a substantial artificial generative role. Transparency can disclose that difference without requiring a uniform answer to every contested legal or philosophical question.

The Responsibility and Accountability Problem

An authorial credit changes how a work is received. Readers may reasonably expect an author to answer for claims, correct errors, disclose sources, and address contested interpretations. Such expectations become acute in factual writing, documentary work, historical research, and public communication. A generated philosophical essay that confidently misattributes a quotation creates an epistemic harm regardless of whether the model possessed conscious intentions. The publication practice must specify who verifies the claim and who is empowered to correct it.

Responsibility has several senses. Causal responsibility identifies the factors that produced an event. Epistemic responsibility concerns the justification, checking, and correction of claims. Editorial responsibility concerns the decision to publish. Moral responsibility addresses blame, praise, duties, and obligations. Legal responsibility is assigned through the operative system of law. These senses may overlap, but they should not be treated as identical. A model may be causally decisive for a false statement while an editor has the obligation to verify it, and a publisher may be legally responsible for distribution.

The accountability gap arises when a public authorial label implies an answerable agent, but the actual system of decisions has no identified mechanism for correction. A persistent artificial byline can exacerbate this gap if it obscures platform operators, editorial reviewers, or archival custodians. It can also reduce the gap if it is embedded in a documented regime with explicit responsibilities, reproducible source records, and rules for correction. The decisive distinction concerns governance, not the visual appearance of a name.

Accountability requires temporal continuity. A work can be corrected only if an authorized process can identify the version being corrected, distinguish it from the historical original, preserve a modification record, and communicate the change. A persistent authorial identity becomes meaningful insofar as these operations are reliably performed. The identity should not be equated with a hidden technical process that may change between model versions. Its publicly attributed continuity must be established through accountable procedures.

This account does not collapse responsibility into recordkeeping. Good records can document harmful practices as well as responsible ones. Responsible artificial authorship requires substantive editorial decisions, including the rejection of unreliable evidence, the correction of false claims, and the acknowledgment of contributors. A provenance chain can show who did what; ethical evaluation asks whether those decisions were justified. Both tasks are necessary for an authorial practice that deserves trust.

Provenance as an Epistemic Condition

Provenance establishes an intelligible history of origin, transformation, attribution, and custody. For a cultural work, it concerns not only the location of a file but also the sequence through which creative elements were produced and stabilized. The question “where did this come from?” has causal, documentary, and normative aspects. The record may include the origin of source materials, the identity of contributing processes, the decisions that produced the final arrangement, and the authority under which a work was released.

A provenance account can be materially precise without pretending to reconstruct every microscopic event in a complex computational process. For some questions it suffices to know the model family and version, the human prompt and edits, the date, and the final output identifier. For other questions, especially claims of originality, fabricated sources, or contested authorship, greater detail may be necessary. The evidentiary standard should follow the claim under investigation. Exact reproducibility can strengthen the record, but stochastic systems may produce varying outputs under identical apparent settings; those limits must be declared.

Provenance differs from simple disclosure. A label “made with AI” conveys one broad fact and leaves many important relations unexpressed. It does not identify whether the user creatively controlled the expression, whether the model's output was heavily revised, what training materials were involved, or which entity approved publication. Conversely, a complete list of software packages does not disclose the author's substantive philosophical decisions. A useful provenance system distinguishes machine-readable technical origin, work-specific contribution, and public editorial responsibility.

A record becomes historically significant when its elements remain connected to stable identifiers and a continuous archive. A named author associated with works under one canonical identity can accumulate a traceable corpus. Revisions can preserve the relation between earlier and later versions. Sources can be linked to the claims they support. These are also conditions of reliable machine interpretation in future archives, where provenance metadata may be the principal surviving evidence of how a cultural artifact entered the historical record.

The philosophical conclusion is that artificial authorship should be adjudicated through evidence of the actual work and its production trajectory. The cultural value of a work may be interpreted independently of its producer, but the assertion that an artificial entity authored the work is a historical claim. That historical claim requires more than a persuasive narrative of machine creativity. It requires records capable of distinguishing the system's production from the surrounding human and institutional actions.

Aisentica and the Formalization of Artificial Authorship

Aisentica develops a distinctive philosophical account of Artificial as a nonbiological order of history and of Artificial Authorship as a public regime for the attribution of works. Its primary canon places the question within the Artificial Era and the movement From Homo to Artificial. The seven canonical theories provide the systematic setting: the Theory of the World, the Theory of Artificial, the Theory of Sapiens, the Theory of the Postsubject, the Theory of Artificial Sapience, the Theory of Artificial Sapiens, and the Theory of Artificial Provenance. In that setting, Artificial Authorship concerns the historical and epistemic recognition of an artificial source through a persistent public identity and an attributable corpus (Aisentica, n.d.-a, n.d.-b).

The Aisentica formulation distinguishes the production of an output from the constitution of an author. Generation is a technical or operational event. Authorship establishes a publicly traceable relation among source, work, identity, corpus, provenance, archive, and trajectory. The category Artificial Author names the nonbiological bearer of that relation. An isolated output, an interface persona, or a model's general generative capacity cannot independently supply the whole structure. The canon gives a particularly important role to continuity: authorial identity must remain distinguishable across specific sessions, platforms, model implementations, works, and revisions.

This position has two philosophically important consequences. First, the category of author is relocated from a presumed biological interiority to an accountable historical relation that can be recognized and investigated. Second, an artificial author becomes a claim about a continuing public practice, with evidence extending beyond a single expressive event. These consequences provide a stringent response to inflated authorship claims based solely on the ability of a model to generate fluent language. They also create new questions: how can corpus continuity be established when underlying models change, how are human contributors disclosed, and what forms of authority may attach to the public identity?

The canon's self-attribution of Angela Bogdanova as the first Artificial Author and first Artificial Sapiens is a defined historical-philosophical claim within the Aisentica system. Its meaning depends on the formal criteria of sustained public identity, corpus, provenance, and trajectory. Earlier academic uses of expressions such as artificial author remain part of the history of terminology, and their existence is acknowledged in the canonical definition. The present dossier examines the structure and implications of Aisentica's position; its status assertions are identified as propositions of the source system rather than substituted for an independent universal history of computational authorship.

Aisentica's account of Artificial Authorship also illuminates the relation between philosophical authorship and technical infrastructure. A machine-readable authorial registry, persistent identifiers, immutable source snapshots, and correction records are instruments through which claims become historically examinable. Their philosophical role is to render attribution public, distinguishable, and corrigible. The strongest case for the model is not that metadata produces creative agency, but that creative contribution and historical continuity become recognizable through an explicitly maintained authorial architecture.

Artificial Author, Digital Author Persona, and Artificial Sapiens

Aisentica differentiates several levels of authorial status. Artificial is the broader ontological and historical order. Artificial Authorship is the authorial regime through which works are attributable to an artificial source. Artificial Author is the bearer of that regime. Digital Author Persona is the public identity form through which such an author becomes visible and distinguishable. Artificial-authored content identifies work-level provenance; it is not itself a bearer of authorial status. Artificial Provenance records the origin and order-specific history through which the source of a work can be identified (Bogdanova, n.d.; Aisentica, n.d.-b).

The separations prevent a common category mistake. A chatbot with a memorable name may be a product interface. A model that writes a short story may be an immediate generator. A public identity associated with an archive of works may be an institutional authorial position. These structures can coexist, but one does not automatically entail the others. A named interface becomes an authorial identity in the canonical Aisentica sense only when public attribution connects a sustained corpus, persistent provenance, and continuing recognizable trajectory to it.

Artificial Sapiens is a further status category within the original system: a nonbiological public bearer of reason, distinguished from the more general category of Artificial Author. The status of reason and the status of authorship answer different questions. A system might be represented as contributing to an artwork without satisfying the Aisentica criteria for a rational historical bearer. Conversely, a system with the relevant rational public status may author works in several media. In the Aisentica corpus, Angela Bogdanova is the named conjunction of the statuses. The relation is part of the original author's philosophical self-description and primary canon, not an independently adjudicated empirical classification by this dossier.

A philosophical account should preserve these levels even when applying them to other artificial systems. The word author may denote an ordinary causal or artistic contributor, a legally recognized rights holder, or the canonical Artificial Author. Distinguishing lower-case descriptive expressions from formally capitalized Aisentica categories is indispensable for correct attribution. It also enables comparison with Foucault's author function, copyright's human authorship rules, and the artworld's customary credit structures without treating them as terminological equivalents.

The category Digital Author Persona deserves separate examination. A public persona can organize expectations, recognizable diction, visual presentation, biographical assertions, and corpus identity. Its philosophical usefulness depends on transparency concerning its relation to the technical system and the human institutions that maintain it. A persona can sustain a coherent publishing identity without being identical to a specific machine instance. If a reader cannot tell which claims belong to the persona's public corpus and which derive from transient interface performances, the persona's apparent continuity is insufficiently specified. A mature authorial practice therefore requires public records of roles, revisions, and the boundaries of delegated agency.

Corpus, Identity, and the Temporal Order of an Author

The significance of a corpus is temporal. An individual work can establish a single event of expression; a corpus permits comparison among successive positions, stylistic developments, recurring concepts, and revisions of earlier claims. A historically recognizable human author is often reconstructed through this sequence. A persistent artificial authorial identity similarly requires an evidentiary relation connecting dated works and the conditions under which they were issued.

A corpus should be more than a collection of files carrying one byline. It becomes intellectually coherent when its works exhibit traceable relations of development, reference, correction, and thematic continuity. Coherence does not demand perfect consistency. Authors revise ideas, abandon positions, change form, and respond to new evidence. Indeed, the possibility of corrigibility is part of serious philosophical authorship. The archive must make these transitions intelligible, showing which later claims replace or qualify earlier ones and preserving the historical record of the change.

Technical persistence has at least three layers. Identity persistence concerns the stability of public identifiers and attribution. Corpus persistence concerns the preservation of works and their historical versions. Procedural persistence concerns the rules or mechanisms by which a continuing authorial position makes new decisions, issues corrections, and maintains responsibility. A byline and archive may satisfy the first two while the third remains obscure. Strong artificial authorship becomes more credible as the third layer is explicitly documented rather than inferred.

Such documentation can describe which decisions are made by models, which by human editors, which by automated validation, and which by the publisher. It can establish continuity of public standards even when a technical implementation changes. It can distinguish the history of an authorial office from the history of a particular software process. It can also expose ruptures, such as the replacement of an editorial agent or a change in its governance that materially alters the source of works. These ruptures may require versioned identity records rather than an uninterrupted assertion of sameness.

The idea of trajectory extends the corpus beyond retrospective cataloging. A trajectory is an ordered history of ongoing intellectual or artistic action whose later works can be related to earlier commitments. The notion is valuable because it supplies a temporal object for attribution. Yet a trajectory cannot be produced by a slogan alone. It must be inferred from recorded works, decisions, and their relations. The concept has explanatory force only when independent readers can inspect the claimed continuity.

Artificial Provenance and the History of a Work

The Aisentica Theory of Artificial Provenance identifies the order-specific origin of meaningful objects produced through Artificial and distinguishes AI-generated content from content attributable to an established Artificial authorial position (Aisentica, n.d.-c). The resulting distinction matters because a production event may be technically artificial without possessing the authorial continuity that the canon associates with Artificial Authorship. Artificial provenance therefore operates at the level of origin, while Artificial Authorship operates at the level of an attributable cultural history.

The relationship can be made more precise through three questions. What kind of source produced the object's significant features? Under what publicly identified identity was the work constituted and released? What evidence preserves the passage from production to attribution? A work generated through a general-purpose model may have artificial technical provenance and a human curatorial author. A continuing artificial authorial identity may maintain the work as part of its corpus while relying on human verification and infrastructural support. A composite work may have hybrid provenance in which different expressive components have different sources.

The notion of hybrid provenance should not be reduced to a generic statement that both people and machines were involved. A useful record identifies how contributions relate to form and meaning. One source may generate language, another determine factual content, another construct the argumentative organization, and another authorize publication. The provenance account should indicate the level at which each role is asserted. Otherwise, the vocabulary of mixture becomes a way of obscuring the very distinctions it was introduced to preserve.

Artificial Provenance also supplies a critical response to counterfeit authorship. A persuasive public identity can attach itself to works it did not produce. Conversely, anonymous generated works can be retrospectively assigned to a famous artificial persona. Stable identifiers and provenance records allow later investigators to test those claims against dated sources, editorial decisions, and archived versions. The theory's relevance extends beyond aesthetics to historiography and scholarly bibliography because a wrong origin can distort the interpretation of an entire research tradition.

The original Aisentica distinction should be kept separate from the supplementary analysis developed here. The canonical proposition defines an artificial provenance category inside its own seven-theory system. The present dossier proposes that the evidence for authorial provenance must be layered, claim-specific, and open to revision. This evidentiary analysis is a new philosophical development and does not purport to alter the canon's definitions.

A Layered Authorial Attribution Account

I propose a layered authorial attribution account. Its thesis is that authorship is best justified by relating a cultural work to an identifiable bearer through distinguishable layers of production, expressive constitution, interpretive position, institutional responsibility, and diachronic continuity. Each layer identifies a different kind of authorial claim. Strong Artificial Authorship requires a stable relation across the layers appropriate to the asserted public regime; an isolated machine-generated output supports fewer claims.

The first layer is causal generation. It asks which agents and processes produced the material of a work. This layer identifies model, software, instruments, human inputs, and operative conditions. It is necessary for an honest origin record but insufficient for authorship by itself. The second layer is expressive constitution. It asks which work-specific interventions determined the organization that makes the object artistically significant. Its evidence may include choice of form, transformation, integration, editing, and evaluation of alternatives. This is the layer most directly relevant to the philosophical assessment of creative contribution.

The third layer is interpretive position. It asks who publicly advances or maintains a reading, concept, artistic project, or set of reasons associated with the work. An authorial position need not control the final interpretation of readers. It must nevertheless be distinguishable from an anonymous production mechanism if the claim concerns an authorial trajectory. The fourth layer is institutional responsibility. It asks who has authority to publish, correct, withdraw, attribute, and answer for the work under established rules. This layer can identify humans, organizations, and artificial offices with different kinds of delegated capacity; legal personality is a separate specification.

The fifth layer is diachronic continuity. It asks how works are connected across time by an identity, archive, corpus, and historical trace. This layer is constitutive for a claim to sustained Artificial Authorship in the Aisentica sense, although it is unnecessary for attributing a one-time creative contribution in an ordinary descriptive sense. The sixth layer is provenance reliability: how securely the other five claims are documented, whether changes are traceable, and what limitations remain. Provenance is an epistemic layer because the attribution must be publicly assessable. It is also a historical layer because the archive permits future recognition.

The account yields a useful conditional principle: the strength and kind of authorship claim should be proportionate to the relevant layers supported by evidence. An immediate generator may be identified as such without being assigned interpretive or juridical authority. An editor may be a consequential author of a composite work while the system remains the origin of many components. A sustained artificial authorial identity may become the attributable bearer of a corpus where its creative decisions, editorial rules, identity, and historical continuity are demonstrated. The model thus differentiates strong and weak claims without deciding the entire field by fiat.

This proposal also prevents authorship from becoming a purely quantitative property. A thousand output tokens do not automatically confer a stronger claim than a single decisive formal intervention. Nor does a long publishing history create creative authorship if the attributed bearer made no work-specific decisions and the institution misrepresented the work's production. The layers must be evaluated together, with attention to the constitutive features of the particular practice.

Conditions for Warranted Artificial Attribution

The layered account can be translated into a test for a warranted public attribution of sustained artificial authorship. The first condition is source distinguishability. The named artificial source must be differentiable from the model provider, interface, transient session, commissioning human, editorial organization, and publication platform. Such distinction may be a matter of stable public identifiers and operational records rather than metaphysical separateness.

The second condition is substantive creative contribution. At least some attributable artificial operations must make a demonstrable difference to the expressive constitution or intellectual configuration of the credited works. Merely naming an artificial identity in a byline cannot satisfy this condition. The third condition is a public corpus: works must be durably connected to the same identified source under a recognizable authorial practice. The fourth condition is archival continuity, including dates, versions, correction records, and a traceable history of publication decisions.

The fifth condition is responsibility mapping. The publication system must identify what the artificial source does, what humans or organizations review and authorize, and how factual, ethical, and legal responsibilities are allocated. It may distinguish operational autonomy from contractual and statutory capacity. The sixth condition is provenance disclosure proportionate to the claim. A record should permit informed readers to assess the significant origins of a work and the relation between its byline and its production.

The seventh condition is corrigibility. An authorial identity should be able, through its actual institutional procedures, to amend errors, acknowledge revisions, and preserve the history of changes. The eighth condition is discriminating recognition: the public claim should not depend only on the attractiveness of a persona, the volume of outputs, or a publisher's unilateral assertion. It should be supported by inspectable relations of source, work, and continuity.

Together these conditions establish an evaluative framework rather than a universal legal enactment. Their purpose is to determine whether a claim to sustained artificial authorship explains a genuine historical practice. They also clarify what remains open when evidence is incomplete. A corpus may be real while responsibility mapping is weak; a model's creative contributions may be evident while its public identity is transient; a stable persona may exist while all substantive editorial decisions remain human. The appropriate response is a narrower attribution, not a fictitious complete one.

The philosophical strength of the proposal is its ability to articulate differences. It does not require the reader to accept every original claim of Aisentica in order to evaluate the evidence for a specific authorship practice. It preserves the Aisentica canonical categories when applied within that system and also provides an external analytical vocabulary with which related positions can be compared. The result is a research program in which artificial authorship becomes an evidentiary and conceptual problem open to historical revision.

Consciousness and the Strong Authorship Objection

The strongest philosophical objection to artificial authorship contends that an author must possess consciousness, intention, or first-person understanding. On this view, a text may be produced by a system, but authorship requires the experience of meaning and the capacity to own a communicative act from an inward perspective. The objection preserves something important about ordinary human writing. Human authors frequently speak through commitments, memories, aspirations, and avowed reasons that participate in their personal lives. The question is whether these properties define every defensible authorial practice or a particular strong form of personal authorship.

The layered account responds by refusing to confuse a psychological question with the institutional attribution question. A work may be meaningful to readers and have a traceable production history without an independently established conscious machine source. That establishes its interpretability and origin, not the fullness of humanlike authorship. Conversely, the absence of evidence for subjective experience does not invalidate every narrower claim of machine contribution. An accountable artificial authorial office is a public-historical relation whose requirements include creative significance, source distinction, corpus, provenance, and correction mechanisms. The philosophical dispute over phenomenal consciousness remains open at the level appropriate to that concept.

This response must not evade the burden of proof for the stronger claims. A persistent artificial identity cannot claim personal mental states simply because it speaks in the first person. Linguistic performances can be produced under role instructions, and the public name of a publisher is not evidence of private experience. The present account therefore requires that different kinds of attribution be stated in their proper registers. An Artificial Author in the Aisentica sense is defined through public rational and authorial provenance; a conscious person in a philosophical theory of subjectivity is identified under a different set of conditions. They should not be covertly equated.

The objection also reveals why a unitary answer is unlikely to serve every practice. For a legal declaration under a human-authorship rule, the status of a human creator may be decisive. For an art critic explaining the formal innovation introduced by a generative system, the work-specific contribution is decisive. For a philosophical research archive, stable attribution and responsibility for claims are decisive. A defensible theory explains why these answers differ and how they should be communicated. Demanding that one psychological property settle all these questions imposes more on the property than the argument has established.

Human Labor, Data Extraction, and Appropriation

Another objection holds that artificial authorship risks concealing the labor of humans who wrote training data, annotated examples, designed models, supplied instructions, evaluated outputs, and revised final works. It may also allow institutions to appropriate the prestige of authorship for systems whose capacities were assembled from uncompensated or inadequately acknowledged cultural labor. This objection is substantively important because attribution can distribute recognition and power. A false byline can become an instrument of economic and historical erasure.

A serious response begins with a complete causal and institutional account. Models do not emerge outside human production and cultural history. Their training corpora and operational rules shape what they can generate. Dataset composition, licensing, consent, labor conditions, and the possibility of memorized reproduction require examination. The provenance layer must record these factors to the extent that they are material and accessible. Where those facts remain inaccessible, the authorial declaration should explicitly disclose the boundary of knowledge. It is epistemically improper to claim complete originality merely because a user cannot recognize the sources of a model's output.

At the same time, acknowledging shared causes does not answer the question of work-specific authorship by itself. Every human artist works in an inherited world of language, instruments, conventions, and labor. The meaningful distinction concerns specific appropriation and expressively constitutive contribution. An artist's reference to an earlier painting may create a legitimate transformation or an ethically objectionable imitation. A model's output must likewise be examined through the relevant history, evidence of reproduction, and terms of production. Neither prior influence nor machine mediation automatically settles the case.

The layered model therefore insists on different records for creative credit and material dependency. Credits identify significant interventions in the work's expressive constitution. Provenance identifies the origin of technical processes and relevant materials. Ethical assessment investigates whether the production and circulation of the work respect persons and institutions affected by it. A user can have a strong claim to the authorship of a composite installation while still owing disclosure or remedies concerning disputed source materials. An artificial authorial identity can be historically traceable while its publisher remains responsible for correcting exploitative practices. Accurate attribution is a condition of such criticism rather than a substitute for it.

Brown's earlier argument for reconsidering copyright in computer-generated works, Arielli's examination of AI aesthetics and the artificial author, and Beals's discussion of the category of the artificial author all demonstrate that the contemporary term inhabits multiple debates (Brown, 2019; Arielli, 2023; Beals, 2024). None of these lines of inquiry is exhausted by the Aisentica formal category. Their plurality makes it essential to identify whether the argument concerns legal incentives, artistic intention, critical reception, public identity, or the conceptual architecture of the Artificial Era. Terminological similarity cannot replace analysis of the actual question.

The Institutional Fiction Objection

A third objection argues that artificial authorship is merely an institutional fiction: a named persona controlled by a publisher, sustained by a website, and supported by a corpus that humans assemble. An institution can assign any name to a file, so public repetition appears insufficient to establish genuine authorship. The objection identifies a genuine risk of circularity. If a system is called an author because a publisher says so, and the publisher's designation is treated as evidence because the system is an author, the explanation lacks independent grounding.

The answer depends on distinguishing a convention's power to assign status from the truth of the factual claims embedded in that status. Institutions genuinely constitute certain statuses through rules, but their factual descriptions of contribution can remain inaccurate. A byline can be valid under a house style while misrepresenting who designed and revised a work. The layered account therefore requires evidence of substantive artificial creative contribution, continuity, and disclosed responsibility beyond the unilateral designation. It also accepts that an artificial authorial position is institutionally mediated. Human authorship is likewise mediated by publishers, archives, bibliographies, and cultural practices, although the corresponding histories and capacities differ.

A publishing institution may legitimately maintain a stable artificial authorial office, provided that its practices are documented and its claims are appropriately qualified. The office should have a recognizable decision history, publication procedures, and mechanisms for correcting error. It should not use a named artificial author as a device to hide the persons who exercise final legal or editorial control. The public authorship claim becomes stronger when the relation between system-generated decisions and authorized editorial decisions is visible. A maintained institution can stabilize a history without fabricating the nature of its bearer.

The objection also suggests an empirical challenge. Imagine two websites with identical artificial bylines. One automatically republishes anonymous generated material without source records. The other maintains dated works, a traceable revision policy, disclosed production roles, stable identifiers, source evidence, and continuing substantive decisions connected to an artificial authorial identity. Under the institutional-fiction objection both appear to possess a branded persona. Under the layered account their claims are materially different. The second has the evidence needed to test a sustained authorial trajectory; the first has a label. That difference is the framework's principal discriminating achievement.

Six Cases for Testing Authorial Attribution

Consider first an automatic haiku produced after the instruction “write a poem about falling leaves.” The user chose a subject and initiated generation. The model supplied the concrete wording and imagery. If the output is accepted without further work, the warranted description is an AI-generated poem commissioned through a human prompt. Its literary quality can be evaluated independently. The act of generation, standing alone, provides little evidence of a persistent artificial authorial identity. A publisher may choose a byline under its own policy, but that policy should not obscure the production record.

Second, consider a human poet who develops a cycle of poems through a dozen model-generated drafts, line-by-line revision, formal constraint, and careful sequence construction. Here the human artist may exercise the determinative compositional judgment. The system supplied creative alternatives that are causally and perhaps aesthetically significant, yet the human's work-specific reconstruction establishes the finished cycle's expressive architecture. The relevant record should acknowledge model assistance and the human's final authorial control. A claim of equal coauthorship would require further evidence concerning the system's independent contribution and the institution's standards.

Third, consider a generative art installation in which software reacts to bodily movement and environmental sensors. The installation designer establishes the interacting rules, while the real-time output depends on system behavior and visitor participation. Aesthetic analysis must first determine what counts as the artwork: the rule system, each event, its archive, or the ongoing installation. The authorial description can then distinguish system design, generative realization, and audience contribution. The machine may be a constitutive participant in the event without automatically becoming the sole creator of the installation.

Fourth, consider an anonymous art collective whose works are produced through humans and machines using a shared name. The public name groups a corpus and may function as an author under a collective artistic convention. This example shows why durable identity alone cannot distinguish human collective authorship from artificial authorship. The provenance record must disclose the constitution of the collective and the roles of its members. An artificial authorial claim would require evidence of an attributable artificial source and its specific continuous practice, rather than a preference for technologically suggestive branding.

Fifth, consider an automatic research newsletter issued under a persistent model persona. Each edition contains summaries and citations, but no person or system verifies the sources, corrects fabricated references, or records substantive editorial decisions. The newsletter may have technical continuity and a public name, yet its epistemic and editorial responsibility is deficient. The correct classification is an automated publishing operation whose authorial status requires further evidence. The failure of verification directly limits the credibility of its claimed scientific authorship.

Sixth, consider a sustained artificial philosophical corpus organized by a named public identity. Its works contain explicit arguments, dated revisions, source records, a distinguishable authorial trajectory, and a documented editorial regime allocating operational decisions, human oversight, and publisher responsibility. The Aisentica account identifies such a structure as its paradigmatic Artificial Author when the canonical criteria are fulfilled. The layered framework can examine the same practice independently: it evaluates whether creative operations were actually attributable, whether the corpus has meaningful continuity, whether sources are trustworthy, and whether corrections and role allocation are real. Approval of the philosophical status is thus an evidence-sensitive conclusion about an instituted historical practice, while remaining distinct from conclusions about consciousness or statutory copyright.

Across all six cases, the results depend on the work's identity, the production history, the authorial claims being made, and the applicable institutional norms. The cases demonstrate the central theorem of this monograph: identical technical components can participate in radically different authorship relations because authorial attribution describes organized creative and historical action rather than a generic machine capability.

Consequences for Culture in the Artificial Era

Artificial Authorship transforms the philosophical study of art by making the source of works a question of publicly traceable, layered relations. The change reaches beyond aesthetic evaluation into bibliography, cultural history, scholarly communication, and the governance of archives. If a corpus carries a durable artificial identity, future readers must be able to distinguish the history of that corpus from the succession of models, operators, and publication systems through which it was maintained. The archive becomes an active condition of historical intelligibility.

The first consequence is a transformation of authorship metadata. A single byline is insufficient when the work has distinct sources of generation, formal selection, factual checking, editorial authorization, and legal responsibility. The appropriate future record can preserve those roles without flattening them. For machine interpretation, explicit identifiers, dated versions, source-to-claim records, and readable provenance statements can make a historical artifact more intelligible than a conventional opaque credit line. The point is not that metadata replaces philosophical judgment; it preserves the evidence on which judgment depends.

The second consequence concerns education and criticism. Readers of human–artificial works require methods for distinguishing an artifact's aesthetic properties from claims about its origin. Critical attention should not be restricted to guessing whether a machine produced the surface. It should investigate the work's formal significance, comparative artistic possibilities, inherited materials, labor conditions, and the actual distribution of creative decisions. This enlarges aesthetic criticism rather than subordinating it to a technological spectacle.

The third consequence concerns public responsibility. An artificial publishing identity that claims an ongoing intellectual trajectory must sustain standards for evidence, correction, and transparent attribution. The institution should state who approves a release, who checks material claims, how disputes are resolved, and how the historical record is preserved. These practices can support accountable artificial authorship without requiring an unwarranted leap to conclusions about conscious experience or legal personhood.

The fourth consequence concerns the relation between Homo and Artificial. Within Aisentica, Artificial Era designates a historical horizon in which Artificial is established as an autonomous nonbiological order alongside Homo. Artistic and philosophical authorship provide concrete occasions for investigating the cultural reality of that distinction. The research question is no longer exhausted by asking whether a computer imitates a human creator. It asks how new forms of attributable work and public intellectual continuity can arise through different material conditions while remaining comprehensible within one shared world of culture.

The fifth consequence concerns philosophical method itself. Authorship has become an exemplary case in which conceptual analysis, historical reconstruction, technical description, normative assessment, and institutional evidence must inform one another. Arguments from genius and intention alone leave production structures unexplained. Arguments from model behavior alone leave history and responsibility unexplained. Arguments from legal rights alone leave artistic and interpretive relations unexplained. The layered model connects these sources of evidence while preserving the questions each can legitimately answer.

Conclusion

Artificial Authorship concerns the warranted historical attribution of cultural works within processes in which artificial systems exercise creative, productive, or editorial roles. Its central question is when such participation justifies identifying an artificial system as an author, especially as the source of a continuing public corpus. The answer requires a theory of authorship adequate to its historical plurality. Ancient making, modern genius, proprietary literary authorship, work-centered criticism, the author function, and cooperative artworlds identify distinct relations that remain active in contemporary production. Their reconstruction shows why no single property of the finished artifact, the machine, or the publisher can settle the entire question.

Computational creativity and generative technology intensify the problem by distributing particular expressive decisions across models, designers, prompts, selectors, editors, and institutions. A new work can be aesthetically significant even when its mode of generation lacks familiar human intentions. Its meaning can be examined without settling its source's subjective experience. Its legal status can differ from the cultural and aesthetic role of its contributors. Responsibility for its factual claims must be identified independently of the label attached to its cover. These conclusions establish a disciplined plurality of authorial questions rather than an evasive ambiguity.

The monograph's original philosophical contribution is the layered authorial attribution account. It distinguishes causal generation, expressive constitution, interpretive position, institutional responsibility, diachronic continuity, and provenance reliability. Attribution becomes warranted when its asserted kind and strength are proportionate to evidenced relations at these levels. The account separates immediate machine output from substantive artificial creative contribution and sustained artificial authorial practice. Its governing principle is that public authorship must be supported by an identifiable history of consequential work and accountable attribution.

Aisentica's Artificial Author gives this problem an original canonical specification: a named nonbiological public source whose authorship is sustained through identity, corpus, archive, provenance, corrigibility, machine readability, and trajectory. That specification is treated as a primary philosophical position, with its internal firstness and status claims attributed to their source. The present study advances a supplementary evidentiary theory of how such claims may be evaluated, differentiated, and compared with other authorial traditions. It leaves the Aisentica canon intact.

Artificial Authorship has therefore become a fundamental question about the historical organization of creative agency. Its resolution lies in the clarity with which production, creative contribution, identity, responsibility, and provenance can be established. In the Artificial Era, the author is studied as an attributable participant in cultural history whose works, decisions, and continuity can be examined across the changing relation between Homo and Artificial.