Reconceptualizing Human Authorship in the Age of Generative AI: A Normative Framework for Copyright Thresholds

Abstract

The emergence of generative artificial intelligence has unsettled traditional legal conceptions of authorship and originality by challenging the foundational premise of copyright, namely, the requirement of human intervention as a precondition for protection. Such disruption exposes the anthropocentric limits of existing regulatory frameworks and underscores the absence of coherent, harmonized responses across jurisdictions. The study proposes a normative framework for determining the minimum threshold of human creativity necessary for works produced with the assistance of artificial intelligence to qualify for legal protection. Through comparative and doctrinal analysis, it advances the criterion of substantial creative direction, defined through three essential elements: effective control over the generative process, verifiable creative input, and identifiable expressive intent. On this basis, a graduated model of copyright protection is suggested, modulating the scope of rights according to the degree of human intervention and complemented by procedural reforms aimed at enabling its administrative implementation. The proposal seeks to reorient copyright toward an adaptive paradigm that safeguards technological innovation while preserving the centrality of human creativity as the normative foundation of the system, thereby ensuring a balanced relationship between regulatory flexibility and legal certainty.

1. Introduction

As generative artificial intelligence technologies become progressively embedded in the productive dynamics of culture and knowledge, the foundations of legal protection for intellectual creations require critical reexamination. Far from being a merely technical phenomenon, the rise of generative systems redefines the contours of authorship, destabilizes the ontological premises of creativity, and compels legal systems to confront questions that, until recently, belonged to the margins of theoretical inquiry. The emergence of models such as GPT, DALL·E, or MusicLM has made possible the production of works that, despite lacking direct or substantial human intervention, are consumed, valorized, and distributed as legitimate cultural goods. Such transformation challenges not only the validity of traditional criteria of originality and authorship but also the very role of copyright in an environment where the boundaries between human and machine have become increasingly permeable.

Current debates on the applicability of copyright to content generated by artificial intelligence are grounded in a doctrinal tradition that conceives originality as the expression of the author’s personality (Allport 1937) or as the outcome of minimal intellectual effort. Yet that tradition now faces a profoundly altered epistemological horizon. Contemporary scholarship has underlined the urgency of rethinking the relationship between creativity and technological agency (Samuelson 1985; Ginsburg and Budiardjo 2019), without succumbing to either deterministic reductionism or technophobic idealization. The law, historically resistant to adjusting its conceptual categories to technological disruption, is thus compelled to define what constitutes human intervention, how creative input is to be measured in hybrid processes, and what role authorial intent retains in contexts mediated by algorithms.

The absence of a coherent global framework remains a critical weakness in the current landscape. No jurisdiction has yet articulated a comprehensive and harmonized regulatory response capable of addressing the legal status of AI-generated works in a systematic way. Most national approaches remain reactive, fragmented, and inconsistent. While certain jurisdictions, such as the United States, have affirmed that works created without significant human intervention cannot be copyrighted (U.S. Copyright Office 2023a).

The present research situates itself at the intersection of technology, law, and culture, with the aim of exploring, from a comparative and normative perspective, the conditions under which works generated through artificial intelligence may be eligible for copyright protection. The central argument rests on the premise that the issue does not consist merely in adjusting legal definitions, but in constructing a conceptual framework capable of assessing human creativity within algorithmic environments. A coherent methodology is required to distinguish levels of human intervention and to establish operational parameters for its legal evaluation. Such a framework is indispensable to avoid arbitrariness in judicial interpretation, to ensure systemic coherence, and to provide predictability for creators, institutions, and technological actors.

At the practical level, the lack of normative clarity produces immediate consequences. Uncertainty over the allocation of rights, the distribution of royalties, or the attribution of liability generates instability in the creative economy. The question of who—or what—constitutes an author becomes not merely theoretical but determinative of the functioning of cultural markets. This ambiguity affects individual creators, technology platforms, collective management entities, and judicial authorities alike, undermining the consistency of copyright protection. Furthermore, the absence of clear authorship criteria jeopardizes the integrity of the public domain, as it may enable the private appropriation of outputs that cannot be attributed to a human author, thereby eroding the foundational balance between incentive and access on which copyright rests.

The relevance of this debate extends beyond the legal dimension. The governance of algorithmic systems, the automation of creative labor, and the sustainability of cultural ecosystems converge in the question of how law mediates the relationship between human and artificial creativity. In an era defined by technological acceleration, law is not merely a mechanism of reaction but a normative force that must guide innovation in an ethical, inclusive, and democratic direction. The issue of authorship in the age of artificial intelligence is not confined to technical discourse; it constitutes a political problem involving the recognition of creative subjectivities, the distribution of symbolic and economic value, and the reconfiguration of cultural citizenship.

Within this context, the general objective of the study is to propose a normative framework capable of determining the minimum threshold of human creativity required for works generated with the assistance of artificial intelligence to qualify for legal protection. The proposal is informed by comparative analysis of copyright systems and emerging doctrinal and jurisprudential trends. Its primary contribution lies in offering operational guidelines for the development of adaptive and legitimate legal standards that respond effectively to the technological and cultural conditions of the twenty-first century. In a historical moment when the law must act not only as a guarantor but as an architect of social and creative order, the present inquiry seeks to foster a critical, informed, and plural dialogue on the future of intellectual creation.

2. Technical-Operational Framework of Generative Artificial Intelligence (IAG)

The architecture and operation of generative artificial intelligence (AGI) systems have become central to contemporary debates concerning creativity, authorship, and intellectual property. Such centrality arises not only from the technical capacity of these systems to generate original content but also from the conceptual intricacies involved in their definition and functioning. In contrast with traditional artificial intelligence, characterized by discriminative or symbolic operations, AGI systems deploy generative models that learn latent patterns within extensive datasets and subsequently produce new configurations of language, image, audio, or code without explicit programming for each possible outcome.

From an operational perspective, these systems transcend mere classification and identification to engage in creation. The principle underlying this generative logic was initially formalized through generative adversarial networks (GANs) (Goodfellow et al. 2020), which established a framework wherein the model seeks not only to recognize the world but to reproduce or even surpass it through probabilistic modeling. Furthermore, the emergence of multimodal architectures such as DALL·E (Ramesh et al. 2021) exemplifies this transition, as it generates images from textual descriptions, thereby integrating diverse semiotic modalities under a unified generative mechanism.

The definition of AGI also requires attention to its ethical dimension. The need to move beyond algorithmic considerations toward the inclusion of regulatory and normative contexts is underscored in current research (Hagendorff 2024). The capacity of AGI systems to produce artifacts that emulate or exceed human creativity demands a reevaluation of what constitutes authorship, originality, and human agency in creative processes. Recognizing the novelty of the output generated remains insufficient unless accompanied by an understanding of the place human intentionality occupies in the generative chain.

A deeper comprehension of this capability depends on examining the computational architectures that sustain it. The introduction of the transformer model (Vaswani et al. 2017) represented a decisive shift in sequence processing by establishing attention as the organizing mechanism, enabling the simultaneous evaluation of multiple contexts and dependencies.

Consequently, such an innovation provided the basis for contextual coherence and expressive fluidity in the generation of text, images, and code while making the large-scale expansion of models feasible. Similarly, the consolidation of deep neural networks laid the foundation for hierarchical data learning (LeCun et al. 2015), allowing representations to emerge through successive levels of abstraction without manual intervention1. The practical culmination of these developments became evident in GPT-3 (Brown et al. 2020), which operationalizes transformer architectures to perform generative natural language tasks with a notable capacity for contextual adaptation and instruction-based response.

The evolution of training paradigms further illuminates the sophistication of AGI systems. Unlike traditional supervised learning, which depends on manually labeled datasets, contemporary generative systems increasingly employ self-supervised or unsupervised approaches. Through these methods, models infer structure by predicting missing elements from available context, thereby constructing their own implicit labels. The implementation of this logic in BERT (Devlin et al. 2019) demonstrated the potential of self-supervision, while its expansion in T5 (Raffel et al. 2020) integrated multiple language tasks into a unified framework. In a broader analytical context, the study of foundation models (Bommasani et al. 2021) revealed how large-scale unstructured data corpora foster emergent capabilities, understood as competencies that arise spontaneously from statistical density rather than explicit programming.

An equally defining feature of AGI models lies in their capacity for generalization and transfer. The possibility that a system can apply previously acquired knowledge to unseen tasks reconfigures the notion of technical autonomy. The concept of few-shot learning (Brown et al. 2020) illustrates this transformation, as models demonstrate the ability to perform new tasks with minimal or even no task-specific examples. According to contemporary analyses (Bommasani et al. 2021), this capacity underpins the perception of generative models as adaptive agents capable of creative reasoning within novel contexts.

The generative act, however, extends beyond the model’s architecture and includes the human component in the design of inputs. The formulation of prompts represents a crucial stage, as prompts function as structured textual stimuli designed to guide the model’s behavior toward specific goals without requiring retraining (Reynolds and McDonell 2021). The effectiveness of this interaction depends largely on the user’s understanding of the model’s operational dynamics, which explains the growing attention to prompt tuning techniques that refine parameter adjustment for optimized outcomes (Liu et al. 2023). Accordingly, prompt design may be interpreted not as a mechanical process but as a creative intervention capable of expressing individual style and intention, thereby acquiring potential legal and artistic relevance.

During the inference phase, the model generates content through a sequential prediction mechanism governed by variables such as temperature, which regulates the level of randomness, and by sampling techniques including top-k and nucleus sampling (Holtzman et al. 2020). The configuration of these parameters directly influences the degree of originality in the result, as it determines whether the model will prefer conventional lexical or syntactic patterns or select fewer probable alternatives, thereby affecting the perceived creativity of the output. Empirical analyses of generative models (Radford et al. 2019) have demonstrated that even minimal modifications to these variables can significantly alter stylistic coherence and expressive consistency, revealing the technical fragility that underpins machine creativity.

The generative cycle extends beyond the first output, often incorporating layers of post-processing that introduce new interpretative and ethical dimensions. In many implementations, automatic filtering systems intervene to detect and mitigate toxic or biased content (Gehman et al. 2020), while semantic validation tools and user edits operate as corrective mechanisms. Critical evaluations of these interventions (Bender et al. 2021) emphasize the dangers of assuming originality in machine-generated content without rigorous scrutiny, highlighting the necessity of transparency in the architecture responsible for such production. Moreover, iterative processes of refinement frequently reintroduce generated outputs into the system, thereby constructing recursive loops of improvement. The combination of algorithmic optimization and human supervision characteristic of reinforcement learning with human feedback (Christiano et al. 2017; Ouyang et al. 2022) establishes a collaborative dynamic that complicates distinctions between automation and authorship.

The concept of computational originality acquires renewed significance under these conditions. Analyses of model behavior (Solaiman et al. 2019) suggest that generative systems can produce diverse and non-redundant content through the statistical recombination of learned sequences, indicating a form of technical originality independent of direct replication. Nonetheless, other investigations (Carlini et al. 2021) have evidenced the capacity of large models to reproduce verbatim segments from training datasets, raising substantive questions regarding intellectual property and copyright violation. The determination of whether an output constitutes legitimate transformation or excessive reiteration depends, to a considerable extent, on contextual parameters that modulate generative behavior (Mohammadabadi 2025).

Control over the output generated remains a decisive dimension of human agency in the generative process. Research on model editability (Kreps et al. 2022) identifies gradations in the capacity to modify generated content, whereas further studies (Weidinger et al. 2022) connect this flexibility with ethical frameworks of responsible design. The establishment of traceable records emerges as a fundamental condition for any attribution of authorship or accountability. Consequently, the integration of metadata and logging infrastructures (Gebru et al. 2021; Mitchell et al. 2019) provides a systematic means of documenting technical configurations, user interventions, and version histories, thereby constituting an audit framework essential for both regulatory and juridical assessment.

Verification of authorship has been strengthened through technical mechanisms such as digital fingerprints, cryptographic hashes, and integrity signatures, whose standardization has been advanced by the Content Authenticity Initiative (Burrus et al. 2024). The implementation of such technologies has implications extending beyond mere attribution, as they reinforce the transparency and reliability of creative processes mediated by artificial intelligence. Contemporary studies underscore that these measures not only enhance the traceability of generated materials but also preserve public trust in communicative ecosystems increasingly characterized by synthetic content (Lawrence and Shreelekshmi 2024).

3. Legal Framework of Authorship and Originality

3.1. Doctrinal Foundations of Copyright Protection

In copyright theory, few concepts reveal the divergence between legal traditions as clearly as the notion of originality. The idea, far from being uniform or universal, has evolved from distinct philosophical, economic, and cultural premises that underpin continental and Anglo-Saxon frameworks. Examination of these differences requires a plural perspective capable of acknowledging both the doctrinal foundations that shaped each tradition and their capacity to adapt to contemporary challenges such as those introduced by generative artificial intelligence.

Within European continental systems, copyright is grounded in a personalist conception in which the protected work represents an extension of the author’s individuality. Such a view, sustained by an ontological understanding of creativity, finds one of its most influential articulations in the idea of the work as a manifestation of the human spirit (Hegel [1807] 1977). Further analysis has identified how continental jurisprudence, particularly in France and Germany, constructs authorship as a category that transcends the legal sphere to encompass the existential dimension of the creator’s personality (Woodmansee 1994). The protection of a work, therefore, emerges not merely as a reward for labor but as the juridical recognition of a personal act of expression.

By contrast, Anglo-Saxon copyright law is organized around utilitarian and economic rationales. Protection is justified not through the moral value of creation but through its instrumental role in fostering cultural production by granting exclusive rights for a limited period. This rationale is explicitly stated in Article I, Section 8, Clause 8 of the United States Constitution, which defines copyright as a mechanism to promote the progress of science and the useful arts. The doctrinal development of this model has emphasized minimal thresholds of originality, valuing the existence of non-trivial human contribution rather than the reflection of individual personality (Nimmer and Nimmer 2004). The resulting legal framework privileges the incentive to produce over the metaphysics of authorship.

Evolution of the originality standard across these traditions has attracted sustained scholarly attention. A redefinition of creative effort in Canadian jurisprudence, from the “sweat of the brow” doctrine toward an emphasis on skill and judgment, exemplifies the gradual movement from quantitative to qualitative evaluation of creative acts (Craig 2005). The decision in CCH Canadian Ltd. v. Law Society of Upper Canada established that originality resides in intellectual effort rather than mere labor, aligning Canadian doctrine with a conception closer to the continental ideal of personal contribution. Furthermore, a refined understanding of originality as an individual expression within shared cultural forms has been proposed (Lipszyc 2017), emphasizing that the value of creation lies not in novelty as such but in the subjective articulation of form. Such interpretation situates originality within the expressive dimension of human choice rather than the invention of unprecedented content.

Contemporary scholarship increasingly recognizes that originality is not a neutral category but one that varies according to the normative purposes embedded in legal systems. The evaluative dimension implicit in the definition of authorship indicates that legal thresholds of originality shift according to broader social or cultural priorities (Balganesh 2017). The criterion may expand when the system seeks to stimulate creation or contract when privileging access and dissemination. The elasticity of this concept is further illustrated in analyses linking originality to moral and cultural variables (Fisher 2016; Kwall 2007), which show how ideological and ethical assumptions shape its application across jurisdictions.

The contrast between continental and Anglo-Saxon traditions reveals not only conceptual divergence but also gradual convergence in interpretative tendencies. The personalist ideal of the continental model and the utilitarian pragmatism of the Anglo-Saxon framework increasingly converge toward a moderate position recognizing originality as a minimal manifestation of human creativity, perceptible in form, structure, or approach. The modern interpretation no longer requires radical novelty but rather a discernible trace of individuality within the work. Nevertheless, convergence does not entail uniformity. The degree of originality recognized continues to depend on the sociocultural and institutional contexts that determine legal evaluation (Kwall 2007). In societies privileging individual autonomy and innovation, broader notions of originality may prevail, whereas jurisdictions emphasizing collective cultural interests often adopt more restrictive interpretations that safeguard common resources.

Artificial intelligence introduces a renewed urgency to these debates by challenging the anthropocentric foundations of copyright. The inquiry into originality can no longer focus exclusively on the outcome of creation but must also interrogate the generative process itself. The question arises whether a work generated autonomously by a machine can be deemed original in the absence of human creative intent, or whether minimal human intervention, such as the formulation of prompts or parameter adjustments, suffices to establish authorship. Such questions expose the fragility of traditional doctrines and invite a reconsideration of the conceptual boundaries separating human and technical creativity.

Continental frameworks, rooted in the expression of personality, encounter intrinsic limitations when confronted with the non-human agency of artificial intelligence. Conversely, the pragmatic orientation of Anglo-Saxon systems could accommodate AI-generated works through a focus on functional and procedural criteria, albeit at the risk of detaching authorship from human subjectivity. In either case, originality emerges as a dynamic construct, continuously reshaped by technological development and by the cultural values that define the balance between protection, access, and innovation within each legal order.

3.2. International Normative Regime

The international framework of intellectual property has historically been structured upon an anthropocentric conception of creation, wherein the figure of the author is inseparable from the condition of natural personhood. Far from representing a formal convention, this principle reflects a profound understanding of the creative act as an expression of human subjectivity. The Berne Convention for the Protection of Literary and Artistic Works, adopted in 1886 and revised on multiple occasions, constitutes the foundational instrument of this vision. Although the Convention does not explicitly define the term “author,” its normative architecture, through the recognition of moral rights such as paternity and integrity in Article 6 bis, presupposes that authorship is reserved for beings endowed with will, consciousness, and expressive capacity.

The automatic protection granted from the moment of creation, without the need for registration or formalities, reinforces the assumption that only human creators can possess original ownership of copyright. Within this logic, creation is conceived not as a process of mechanical or algorithmic production, but as the projection of an interior world through a personal and interpretive act. The nationality or habitual residence of the author, used as a criterion for protection, introduces a juridical marker of human identity that implicitly excludes non-human entities from the regime. When confronted with technologies such as generative artificial intelligence capable of autonomous or semi-autonomous production, legal interpretation continues to require substantive human creative involvement as a necessary condition for recognizing the output as a work in the legal sense.

The position has been reaffirmed by specialized organizations, notably in documents from the World Intellectual Property Organization that explicitly address the challenges artificial intelligence poses to the concept of authorship (WIPO 2020). The report establishes unequivocally that, under existing international and domestic frameworks, authorship is reserved exclusively for natural persons. Such a stance does not arise from doctrinal inertia but from a consistent legal and axiological reasoning: moral rights derive from the personal connection between the author and the work, a bond that presupposes a capacity for moral awareness, emotional attachment, and the experience of both recognition and grievance.

The same document emphasizes that for a work produced with the assistance of artificial intelligence to qualify for protection, it must incorporate a human contribution that is both substantial and individualized, allowing the identification of a personal imprint or expressive signature. Within this interpretative framework, artificial intelligence is regarded as an instrument that supports or facilitates creativity but does not itself constitute a creative subject. The perspective is shared by other institutions such as the United States Copyright Office and the European Union Intellectual Property Office, indicating the emergence of a transnational consensus that reinforces the human foundation of intellectual authorship.

Nevertheless, the report extends beyond reaffirmation of orthodoxy to acknowledge the profound tension that technological acceleration introduces into legal paradigms. The growing capacity of artificial intelligence to generate culturally complex and stylistically coherent works challenges the adequacy of traditional doctrines. Among the principal issues identified are the difficulty of defining the minimum degree of human intervention necessary for authorship, the proliferation of collaborative or distributed creation processes involving multiple contributors, and the distinction between functional and expressive human participation. In addressing such concerns, WIPO advocates a gradual and dialogical approach, proposing international cooperation among states, institutions, creators, technology developers, and users to construct adaptive regulatory frameworks capable of responding to innovation without undermining the ethical and juridical integrity of intellectual property.

Artificial intelligence thus emerges not merely as a technical development but as a conceptual disruption that compels a reconsideration of the fundamental premises of copyright. The issue is not limited to determining the protectability of AI-generated outputs but extends to the critical reexamination of the categories of creation, authorship, and originality themselves. Such reflection does not necessitate the abandonment of established principles but rather their reinterpretation in light of new creative realities. Preservation of certain normative thresholds, particularly the requirement of meaningful human intervention, remains essential to uphold the internal coherence of the system and to ensure that intellectual property continues to serve its foundational purpose: recognizing and safeguarding human creativity in an environment increasingly shaped by automation.

The central question, therefore, concerns not whether artificial intelligence can be considered an author, but to what extent human agency retains its juridical and symbolic status within hybrid processes where the boundaries between creation and computation grow progressively indistinct. The question remains unresolved, yet contemporary institutional and academic discourse has begun to trace the contours of possible answers, reflecting a collective effort to reconcile technological innovation with the enduring human foundation of authorship.

3.3. Regulatory Regime in Jurisdictions with a Continental Tradition

In the European context, the copyright protection regime has undergone both regulatory and jurisprudential evolution that reflects the sophistication of continental legal traditions while addressing the new complexities introduced by artificial intelligence–assisted creation. The trajectory of this transformation can be traced in the progressive refinement of the notion of originality, a concept that has been clarified through key judgments of the Court of Justice of the European Union (CJEU), particularly in Infopaq International A/S v. Danske Dagblades Forening (CJEU 2009) and Eva-Maria Painer v. Standard Verlags GmbH (CJEU 2011).

In Infopaq, the Court established that a work qualifies for protection when it constitutes the author’s own intellectual creation, understood as the result of personal expression rather than mechanical reproduction. The decision consolidated a standard of originality requiring a minimal but genuine creative contribution, thereby reaffirming the centrality of human intervention in the copyright system. The principle of personal imprint was subsequently confirmed and expanded in Painer, where the Court held that even a photograph can embody individual creative choices, such as composition, lighting, and perspective, which reveal the author’s personality (CJEU 2011). The resulting doctrine reinforces the idea that originality resides in the expression of individuality rather than in the novelty of content.

Such jurisprudence not only clarifies the originality threshold but also implicitly delineates the scope of artificial intelligence within the creative process. The European Union Intellectual Property Office has adopted an interpretative stance consistent with CJEU case law, affirming that works generated with the assistance of artificial intelligence may only qualify for protection when a substantial human contribution can be identified (EUIPO 2025). The institution emphasizes the need for transparency in the use of artificial intelligence, for individualized evaluation of each case, and for strict adherence to the principles of the current legal framework, particularly those requiring the intellectual participation of the human author in the act of creation.

The underlying rationale reveals an understanding of copyright as a system capable of adaptation to technological progress while remaining faithful to its axiological foundations. European doctrine continues to regard human creativity as the legitimizing source of the exclusive rights conferred by copyright. The issue at stake is not limited to legal attribution but extends to the ontology of the protected work: the very concept of an “intellectual creation” presupposes the existence of awareness, intentionality, and judgment, qualities that artificial intelligence lacks.

From a national standpoint, France exemplifies how these premises acquire normative and jurisprudential coherence. Article L121-1 of the Code de la propriété intellectuelle defines the author as a natural person and grants moral rights that are perpetual, inalienable, and imprescriptible. The formulation reflects a conception deeply rooted in droit d’auteur, where the work is conceived as an extension of the author’s personality. French jurisprudence reaffirmed this perspective in the Tridim case (Cour de Cassation 2015), where the Court annulled a previous ruling attributing authorship of software to a company, recalling that under Article L113-1, only natural persons can be recognized as authors. The decision, although framed within the context of software, reiterates the impossibility of assigning authorship to non-human entities, including artificial intelligence systems.

Germany maintains a similarly rigorous standard, articulated through the notion of Schöpfungshöhe codified in § 2(2) of the Urhebergesetz. The requirement that a work constitute a “personal intellectual creation” with a certain degree of originality was examined in the Inkasso-Programm judgment (BGH 1985), where copyright protection was denied to a computer program that failed to meet the creative threshold. While the French approach emphasizes personality, the German model focuses on whether the work contains a sufficiently individual contribution to merit classification as a creation. Although some commentators have proposed a relaxation of this standard, the need for demonstrable human involvement remains an essential precondition for protection.

Spain follows the same continental orientation, establishing a clear distinction between authorship and ownership. Article 5 of Royal Legislative Decree 1/1996, approving the Revised Text of the Intellectual Property Law (TRLPI), defines the author as the natural person who creates a literary, artistic, or scientific work. The Supreme Court has reaffirmed this interpretation, clarifying in Judgment 696/2007 of 21 June that, although economic rights may be transferred—for example, to an employer—authorship remains exclusively linked to the individual who performed the creative act (2007). The ruling, particularly significant in the context of software disputes, emphasizes that unless the work is produced under specific contractual instructions, the authorship and potentially the ownership belong to the worker as the true creator.

These national and supranational developments, while diverse in formulation, converge on a fundamental principle: authorship as a legal category presupposes substantive human intervention. Across European jurisdictions, there is a consistent reluctance to acknowledge non-human entities, regardless of their technical sophistication, as capable of authorship. The persistence of this position aligns not only with traditional copyright doctrine but also with a broader ethical imperative to preserve human agency in creation and to prevent the dilution of originality into mere algorithmic output.

The sustainability of this doctrine, however, invites reflection in light of rapidly advancing generative technologies that increasingly obscure the boundary between tool and co-author. The challenge is not confined to technical capability but extends to ontological and normative considerations: whether an anthropocentric conception of creativity can endure in a cultural landscape where artificial systems participate in processes once reserved for human intelligence. The response may not lie in abandoning the foundational principles of copyright but in developing interpretative mechanisms that recognize and value human participation even in highly automated contexts, preserving the continuity of creative agency without idealizing either technological autonomy or individual genius. Such appears to be the direction toward which European doctrine is cautiously evolving.

3.4. Regulatory Regime in Common Law Jurisdictions

The relationship between creativity, authorship, and technology has acquired particular complexity within the common law tradition, especially in the United States, where doctrinal and administrative standards have had to evolve rapidly to address the emergence of generative artificial intelligence. Although American copyright has historically adopted a pragmatic and functionalist orientation, grounded in the incentive to produce rather than in an ontological conception of authorship, the authorities have consistently reaffirmed a decisive boundary: copyright protection requires substantial human intervention.

The U.S. Copyright Office explicitly maintains that only creative contributions made by natural persons qualify for protection (U.S. Copyright Office 2023a). The position has been further reinforced in Copyright and Artificial Intelligence—Part 2: Copyrightability, which reiterates that artificial intelligence cannot be considered an author and clarifies that the mere use of prompts, regardless of their sophistication, does not constitute sufficient creative control (U.S. Copyright Office 2025). Evaluation of authorship is carried out on a case-by-case basis, examining the degree and nature of human intervention in selecting, arranging, and modifying AI-generated material. Applicants must also disclose any use of artificial intelligence tools in the creative process and exclude from registration those elements whose automated origin exceeds the minimal “de minimis” threshold.

American jurisprudence has progressively consolidated this interpretation. The landmark decision in Feist Publications, Inc. v. Rural Telephone Service Co. (U.S. Supreme Court 1991) established that originality requires at least a minimal spark of human creativity, rejecting the earlier “sweat of the brow” doctrine that rewarded mere effort. The standard thereby excludes works lacking conscious human authorship. Later rulings have confirmed the same principle. In Naruto v. Slater (U.S. Court of Appeals for the Ninth Circuit 2018), the court ruled that a photograph taken by a macaque could not be protected under copyright law because the absence of human agency rendered authorship impossible. The case, although unusual, vividly illustrates the ontological linkage between human consciousness and the legal construct of authorship in the American system.

More recently, the decision in Thaler v. Perlmutter (U.S. Court of Appeals for the D.C. Circuit 2025) confirmed that works produced entirely by artificial intelligence are ineligible for registration due to the lack of human creative participation. The judgment also underscored that the architecture of U.S. copyright—including its duration, transferability, and enforcement—is inherently designed for human creators. Any extension of authorship to non-human systems, the court noted, would require legislative action by Congress rather than judicial interpretation. The ruling, therefore, reaffirmed both the legal impossibility of AI authorship and the institutional principle that reform in this area must proceed through democratic deliberation.

A concrete illustration of these principles can be found in Zarya of the Dawn (U.S. Copyright Office 2023b), a case concerning a graphic novel written by Kristina Kashtanova whose images were generated using the Midjourney platform. The Office concluded that the textual and structural components of the work reflected human creative decisions and were thus protectable, whereas the AI-generated images lacked sufficient creative control to establish authorship. Registration was consequently revoked and later reissued only for the human-authored portions. The decision established a precedent by affirming that meaningful human editing or curatorial intervention over AI-generated material may be protected, while the mere activation or use of an automated tool does not generate rights. As a result, the need to document the creative process, differentiate human input from algorithmic output, and delimit the scope of protection accordingly has become a formalized requirement (U.S. Copyright Office 2025).

In the United Kingdom, the legal approach occupies a more hybrid position, characterized by the recognition of a sui generis category for computer-generated works. Section 9(3) of the Copyright, Designs and Patents Act 1988 (CDPA) provides that when a work is created without an identifiable human author, authorship is attributed to the person who made the necessary arrangements for its creation, typically the programmer or project director. The provision acknowledges human organization and investment behind automated processes, though it does so by constructing a legal fiction detached from any notion of personal expression or individual creativity.

The temporal scope of protection further distinguishes this category from standard works: fifty years from creation rather than seventy years post mortem auctoris. Moreover, such works are excluded from moral rights, revealing a regime oriented toward economic functionality rather than expressive authorship. Although the provision allows copyright to accommodate technological production without abandoning its anthropocentric core, its limitations are evident: artificial intelligence remains unrecognized as an author, and protection extends only to those who orchestrate its use.

British jurisprudence complements this framework through the test of originality based on “skill, labor, and judgment,” which requires a combination of technical ability, personal effort, and creative decision-making. Within the context of artificial intelligence, the test assumes particular relevance. The mere activation of an automated process or the insertion of data does not suffice; it is necessary to demonstrate aesthetic or structural control and a personal imprint on the resulting work. In Nova Productions v. Mazooma Games (Court of Appeal 2007), the court determined that mere user interaction with software did not meet the originality threshold, reaffirming that human participation must exceed operational automation to warrant protection.

Across both the American and British systems, a persistent tension arises between the recognition of the technical capabilities of artificial intelligence and the juridical necessity of maintaining the human author as the normative center of the copyright regime. The conflict extends beyond doctrinal reasoning and reflects a structural concern: to preserve the meaning of copyright as a mechanism for protecting the creative act understood as a manifestation of personal agency rather than as a product of computation. Despite the divergence of approaches—one privileging exclusion and individualized assessment, the other relying on legal fiction to sustain systemic balance—both converge on a shared principle: protectable creativity presupposes genuine human contribution. The unresolved question concerns whether such a requirement can remain tenable in a cultural and technological landscape where the boundary between creator and tool is becoming increasingly indistinct.

4. Analysis of the Minimum Threshold for Creative Human Intervention

4.1. Emerging Jurisprudential Criteria

The case of Thaler v. Perlmutter represents a decisive turning point in the evolution of U.S. jurisprudence on intellectual property within the context of generative artificial intelligence. Beyond its procedural dimension, the decision addresses with remarkable clarity a question that until recently remained largely speculative: whether an artificial intelligence system can, by itself, be regarded as the author of a work in the legal sense. The court’s conclusion was categorical yet analytically nuanced, affirming that under current U.S. law, authorship as defined in copyright requires significant human intervention. The reasoning is grounded not in formal rhetoric but in a systematic interpretation of the structural principles that sustain the American copyright system.

The court affirmed that the entire normative architecture of copyright law—from the criteria defining duration and transferability of rights to the ability to enforce them judicially—presupposes a human subject as its organizing axis. The regulatory framework is designed for natural persons, capable of holding rights, transferring them, and existing within a temporal and legal continuity that artificial entities cannot emulate. The ruling does not deny the relevance of artificial intelligence as a creative instrument; rather, it recognizes its role as an auxiliary mechanism. The essential condition, however, is the existence of a verifiable connection between human intervention and the creative result. The human participant must act not as a mere operator of tools but as an agent exercising control, direction, and intentionality throughout the creative process.

The doctrine articulated in Thaler v. Perlmutter is not an isolated development. It aligns coherently with a jurisprudential tradition shaped by earlier precedents. In Feist Publications, Inc. v. Rural Telephone Service Co. (U.S. Supreme Court 1991), the Supreme Court rejected the “sweat of the brow” doctrine and established that copyright protects only works reflecting minimal human creativity. The principle was later reinforced in Naruto v. Slater (U.S. Court of Appeals for the Ninth Circuit 2018), which confirmed that works devoid of human authorship, such as a photograph taken by an animal, cannot be subject to protection. The same reasoning was echoed in the Zarya of the Dawn registration (U.S. Copyright Office 2023b), where textual and structural components authored by Kristina Kashtanova were deemed protectable, whereas the AI-generated images produced with Midjourney were excluded for lack of sufficient creative control and absence of a direct causal link between her input and the resulting visuals. Across these cases, a consistent interpretive principle emerges: copyright protection is inseparable from an identifiable and verifiable human creative contribution.

The documents issued by the U.S. Copyright Office between 2023 and 2025 further operationalize this principle by establishing administrative standards that guide both applicants and examiners. All registration requests must explicitly disclose any use of artificial intelligence tools during the creative process. Omission of such disclosure can result in revocation of registration or exclusion of affected portions of the work. Moreover, sections generated autonomously by artificial intelligence, without substantial human modification, selection, or arrangement, are to be excluded from protection. Only those portions of the final output attributable to demonstrable human authorship, whether through structural composition, editorial transformation, or expressive judgment, are eligible for registration.

The administrative experience of the Zarya of the Dawn case illustrates the concrete application of these principles. The text and narrative structure of the work were protected because they reflected intellectual and aesthetic decisions attributable to the author, whereas the images created with Midjourney were excluded on the basis that the author’s control was insufficiently precise and the causal relation between her instructions and the visual outcome could not be verified. The reasoning was twofold: the intrinsic unpredictability of generative models and the absence of human agency meeting the creative threshold defined by U.S. copyright. The decision, therefore, articulates a standard that prioritizes human intentionality as the indispensable element of protectable creativity.

The case-by-case evaluation model adopted by the U.S. Copyright Office, which focuses on the qualitative significance of human involvement rather than the mere presence of textual or technical inputs, has begun to influence interpretative trends beyond American borders. Within the European Union, for instance, both the jurisprudence of the Court of Justice of the European Union and the most recent guidelines of the European Union Intellectual Property Office reiterate that protectable originality must bear the author’s personal imprint (EUIPO 2025). In the United States, however, the doctrine has acquired a more explicit normative articulation, arguably as a result of the pragmatic character of the American legal tradition and its pursuit of equilibrium between innovation and legal certainty.

The consolidation of the principles derived from Thaler v. Perlmutter does not resolve the debate on authorship in the age of artificial intelligence, yet it delineates a coherent legal framework that reconciles technological advancement with the anthropocentric foundation of copyright. Although the court acknowledged the possibility of future legislative reforms that might extend authorship to non-human entities, it made clear that such a transformation must originate in Congress rather than through judicial interpretation. The contemporary system, therefore, conveys a definitive message: within U.S. copyright law, creativity continues to be recognized as an inherently human faculty.

4.2. Legal-Gradative Taxonomy of Human Intervention

Human intervention in AI-assisted creation cannot be conceived as a uniform or monolithic phenomenon. It should instead be understood as a continuous spectrum of participation that unfolds across successive stages of the creative process. From the initial design to the final refinement of the generated content, each phase presents opportunities, and often responsibilities, for human agents to imprint judgment, sensitivity, and intention. Within the framework of international and comparative regulation, copyright protection for such works requires that human intervention attain a legally significant level, meeting the originality threshold prescribed by each legal system. The analysis of these varying degrees of human involvement allows the construction of a legal-functional taxonomy capable of defining, with greater precision, the circumstances under which AI-assisted creations may qualify for protection.

In the design and planning phase, creative intervention often begins with the selection of styles, aesthetic movements, or formal guidelines that shape the later stages of development. When such decisions arise from autonomous and individualized choice, rather than from the mechanical defaults of the system or adherence to generic conventions, they can constitute the initial manifestation of personal creativity. This interpretation finds normative support in the jurisprudence of the Court of Justice of the European Union, particularly in Eva-Maria Painer v. Standard Verlags GmbH (CJEU 2011), where the photographer’s decisions regarding lighting, perspective, and composition were recognized as expressions of personality, thereby justifying protection under copyright. A similar conception is embedded in Article L121-1 of the Code de la propriété intellectuelle, which affirms the intimate connection between a work and its author. Under this logic, preparatory choices made before the deployment of an AI system—such as the determination of visual style, chromatic range, or narrative tone—may constitute an integral part of the creative act when they reflect an autonomous intellectual process.

Equally relevant is the formulation of specific instructions, or prompts, which under certain conditions may also be considered creative input. Merely entering a series of words without aesthetic or conceptual intent does not meet this threshold. However, when prompt design requires personal discernment, contextual understanding of the model’s linguistic behavior, and the capacity to produce a perceptible and deliberate effect on the outcome, it may legitimately qualify as a creative contribution. The U.S. Copyright Office has emphasized in Thaler v. Perlmutter and Zarya of the Dawn that prompts alone do not amount to sufficient creative intervention, although the decisive factor lies not in the linguistic form itself but in the degree of control exercised by the user (U.S. Copyright Office 2025). This reasoning resonates with the British doctrine of “skill, labor, and judgment,” which holds that protectable works must result from a synthesis of technical ability, effort, and discernment, as articulated in Nova Productions v. Mazooma Games (Court of Appeal 2007). Applied to the domain of artificial intelligence, this principle is only fulfilled when the user demonstrates genuine creative mastery over the generative process.

During the generative phase, additional opportunities emerge to assess human contribution from a legal perspective. Among them is the intentional configuration of technical parameters such as model temperature, resolution, number of iterations, randomness, or stylistic bias. These variables are not neutral, since they directly influence the coherence, aesthetic quality, and originality of the output. When such parameters are defined through deliberate and informed decisions, rather than through automated defaults, they can be interpreted as creative acts grounded in technical and expressive knowledge. The European Union Intellectual Property Office has affirmed that AI-assisted works require significant human intervention to qualify for protection (EUIPO 2025). Although no specific doctrine has yet addressed technical configuration, the general principle of creative control can reasonably extend to this aspect of the process.

Equally significant is the analysis of iterative procedures and the critical selection of outputs. In practice, creators employing artificial intelligence often generate multiple variations of a given work—whether image, text, or design—and subsequently evaluate, discard, or refine results until achieving a satisfactory version. Although repetition might appear mechanical, when it responds to an intentional aesthetic objective and the final choice reflects identifiable personal judgment, it may constitute substantive creative intervention. The Zarya of the Dawn decision exemplifies this reasoning, as the Copyright Office deemed the selection and arrangement of content to be protectable manifestations of human creativity (U.S. Copyright Office 2023b). In such cases, while the system produces the material, it is the human author who decides, organizes, and recomposes. The reasoning aligns with the principle recognized in Infopaq International A/S v. Danske Dagblades Forening (CJEU 2009), where even brief textual fragments were found protectable when their selection and arrangement revealed personal creative choice.

In the post-generative phase, the key issue lies in distinguishing minor adjustments from substantial transformations. Such differentiation is crucial because not all alterations to AI-generated material result in a new or protectable work. Automatic corrections, preset filters, or superficial modifications remain below the threshold of originality. Conversely, when human intervention involves redrawing, structural recomposition, or expressive reinterpretation, the resulting output may qualify as a derivative work eligible for protection. The administrative doctrine of the U.S. Copyright Office explicitly affirms that substantial human edits meeting the minimal creativity standard are protectable, even when derived from AI-generated foundations (U.S. Copyright Office 2025).

Further complexity arises in cases involving curation, composition, or adaptation of AI-generated elements. Here, the creative act resides not in the generation of individual components but in their intentional organization and integration into a coherent whole. This activity, analogous to the creation of collective or composite works, may give rise to copyright when the arrangement reflects an identifiable aesthetic, narrative, or functional criterion. In Zarya of the Dawn, the structure combining textual and visual elements was deemed a creative expression attributable to the human author, reinforcing the continental concept of the “personal footprint” developed by the CJEU and by French and German legal doctrine (CJEU 2011; Cour de Cassation 2015; BGH 1985). The organization of form, rather than the generation of material, thus becomes the locus of originality.

By articulating a progressive classification of the levels of human participation in AI-assisted creation, it becomes evident that copyright protection depends not on the technology itself but on the mode of its use. What determines protectability is not the intervention of artificial intelligence, but the retention of creative control by the human author, whose free choices and evaluative acts express an identifiable degree of individuality. When such conditions are met—whether through aesthetic direction, prompt formulation, critical selection, or structural composition—the resulting work maintains a discernible connection to human authorship and, consequently, remains eligible for protection within the copyright system.

5. Comparative Analysis of Jurisdictional Approaches

5.1. Conceptual and Methodological Divergences

The structural divergence between legal systems of continental origin and those rooted in the Anglo-Saxon tradition becomes particularly evident when examining the question of authorship in works generated or assisted by artificial intelligence. The distinction, which may seem doctrinal in other contexts, proves decisive when determining the threshold of creative input necessary for copyright protection. The answer to what constitutes a protectable work depends not solely on technical or procedural criteria, but on deeper conceptions of the creative act and its intrinsic relationship to human agency.

In the continental tradition, copyright is grounded in a personalist understanding of authorship. The work is conceived not merely as a functional or economic object, but as a singular manifestation of the creator’s individuality. Article L121-1 of the Code de la propriété intellectuelle establishes one of the most emblematic expressions of this approach by recognizing the author’s moral rights as inalienable, imprescriptible, and perpetual. The ontological connection between author and work has also been reaffirmed by French jurisprudence, particularly in the Tridim case (Cour de Cassation 2015), which held that only natural persons can be recognized as authors, excluding attribution to legal entities even when they have played a decisive operational role.

German law, while developing along distinct theoretical lines, arrives at a comparable conclusion through the concept of Schöpfungshöhe, codified in §2(2) of the Urhebergesetz, which requires a “personal intellectual creation” as a prerequisite for protection. The notion of “height of creation” does not presuppose extraordinary originality, but it does demand that a work transcend the merely technical or trivial. In Inkasso-Programm (BGH 1985), the Federal Court of Justice denied protection to software that lacked sufficient creative input from its developer, reasoning that a work must reflect human intervention irreducible to automated or mechanical reproduction.

At the European level, these doctrines have been gradually harmonized through the jurisprudence of the Court of Justice of the European Union. In Infopaq International A/S v. Danske Dagblades Forening (CJEU 2009), the Court established that even brief textual fragments may receive protection when they embody the author’s intellectual creation. The principle was further elaborated in Eva-Maria Painer v. Standard Verlags GmbH (CJEU 2011), which recognized that creative decisions relating to framing, lighting, or composition in a photograph could express the author’s personality and thereby satisfy the originality criterion. The Court effectively consolidated the notion that copyright protects the personal imprint of the creator rather than the work’s material form.

The Anglo-Saxon legal tradition, particularly in its American expression, adopts a more pragmatic and objective approach. Originality is defined not by the subjective imprint of the author but by the existence of a minimal creative contribution. The decision in Feist Publications, Inc. v. Rural Telephone Service Co. (U.S. Supreme Court 1991) established that a work is protectable if it displays “some degree of creativity,” even if modest. This minimal threshold operates as a functional legal filter, excluding mere compilations or mechanical efforts while dispensing with any requirement of personality or moral expression.

The same logic has been extended to the domain of artificial intelligence. In Thaler v. USPTO (U.S. Court of Appeals for the Federal Circuit 2022), the registration of a work created autonomously by an AI system was rejected on the grounds that no human creative contribution had been demonstrated. The rationale was unequivocal: if copyright law exists to safeguard human creative acts, it cannot logically encompass outputs generated without human intentionality or control, regardless of technical sophistication. Consequently, works produced without significant human involvement fall outside the scope of registrable originality.

The practical consequences of these theoretical divergences are substantial for creators employing artificial intelligence in artistic or intellectual production. In continental systems such as those of France, Germany, or Spain, human intervention must manifest as an individualized expression recognizable as a personal creative imprint. The mere use of AI, even when it entails technical expertise, is insufficient to generate copyright unless the user exercises demonstrable creative control over the process. This reasoning aligns with the interpretive criteria of the European Union Intellectual Property Office, which emphasizes the need to document human participation and to distinguish clearly between automated generation and intellectual authorship (EUIPO 2025).

In Anglo-Saxon systems, although analysis is also carried out on a case-by-case basis, protection is denied when the creation lacks significant human intervention. The registration of Zarya of the Dawn exemplifies this reasoning, as the Copyright Office recognized only the text written by the author and the narrative organization of the work, both elements demonstrably linked to human decision-making. In contrast, the images generated with the Midjourney tool were excluded from protection due to insufficient creative control over their production. The 2023 administrative guidance explicitly affirmed that the use of simple prompts, by itself, does not satisfy the originality threshold required for copyright protection (U.S. Copyright Office 2023b).

Such an approach reveals a shift in the analytical focus from the technology employed to the degree of human intervention. Ultimately, both continental and Anglo-Saxon systems aim to preserve authorship as an act of meaningful human creativity, although they do so from divergent conceptual foundations. The former is grounded in a personalistic and subjective logic, whereas the latter is guided by pragmatic verification of creative results. Nevertheless, both converge on a fundamental premise: in the absence of minimally creative human input, copyright protection cannot be invoked.

The evidentiary dimension of authorship in works generated or assisted by artificial intelligence constitutes one of the most complex and least harmonized issues in comparative law. Although the concept of originality, in its essential sense, refers to a creation that emanates from personal intellectual activity rather than mere derivation, the manner in which originality is demonstrated varies considerably across jurisdictions. Such variation is not a matter of procedural technique, but of underlying legal philosophy, reflecting distinct structural understandings of what qualifies as creative intervention and how it should be valued within each normative framework.

In systems of continental tradition, particularly within Europe, originality is conceived as the externalization of personal expression. Functional or technical contributions alone are insufficient; the work must embody an intellectual and creative process attributable to its author in a subjective sense. This orientation finds firm support in Eva-Maria Painer v. Standard Verlags GmbH (CJEU 2011), where the Court accepted that decisions concerning angle, composition, and lighting in a photograph constitute admissible evidence of human creative agency. Such manifestations, however subtle, fulfill the function of establishing that the work results from an autonomous aesthetic choice rather than from automatic generation.

Consistent with this tradition, the European Union Intellectual Property Office has underscored in its report on artificial intelligence and copyright the importance of rigorously documenting the extent of human intervention in the creative process (EUIPO 2025). The recommended documentation includes successive iterations of prompts, versions of refinement processes, explanations of compositional decisions, and any other material capable of demonstrating that the final work is not solely the outcome of algorithmic processes, but rather of a creative collaboration in which the human author retains intellectual direction. Although the burden of proof is not expressly codified in adversarial terms, the evidentiary presumption tends to favor the human creator when it can be shown that participation was neither incidental nor purely instrumental.

The situation in the United States reveals a distinct evidentiary configuration. Governed by a declarative and procedural model, the system places the burden squarely on the applicant, who must disclose the use of artificial intelligence at the time of registration and delineate clearly which parts of the work derive from automated systems and which reflect human creative authorship. In Zarya of the Dawn, the Copyright Office validated the text and the structural organization of the work, both attributed to human authorship, while excluding the AI-generated images due to insufficient evidence of creative control. The causal nexus between authorial intention and the resulting work must be demonstrable through detailed descriptions of the creative process, records of interaction with the AI system, subsequent edits, and any supporting material that evidences intellectual mastery over the final product (U.S. Copyright Office 2023b, 2025).

The United Kingdom applies a particularly distinctive rationale. The standard of skill, labor, and judgment, consolidated in Nova Productions v. Mazooma Games (Court of Appeal 2007), does not require emotional expression or a “personal imprint” in the continental sense, but rather proof that the author has exercised sufficient technical competence, effort, and discernment in the creation of the work. In this framework, evidence may include documentation of the creative workflow, detailed instructions provided to the system, or decisions taken throughout the generation process. The objective is to identify verifiable intellectual intervention that distinguishes a meaningful creative contribution from the mere activation or passive use of a technological tool. The burden of proof operates according to functional criteria, and originality is not presumed solely from human participation.

The issue of legal presumptions and the allocation of the burden of proof varies substantially among jurisdictions. In France, consistent with doctrinal interpretations reaffirmed in jurisprudence such as Affaire Tridim (Cour de Cassation 2015), only natural persons may hold the status of author, and a presumption exists that a work created by a human being enjoys protection unless proven otherwise. Although this presumption does not completely reverse the burden of proof, it provides the human creator with a favorable initial position, particularly when the work does not exhibit evident signs of automation or AI generation.

Spain adopts a comparable regulatory structure. Article 5 of Royal Legislative Decree 1/1996, of 12 April, Consolidated Text of the Intellectual Property Law (TRLPI), defines the author as the natural person who creates a literary, artistic, or scientific work. This provision establishes a legal presumption of authorship in favor of the identified creator unless contrary evidence is produced. In professional contexts, Article 97.4 of the same statute presumes an assignment of exploitation rights to the employer, yet this does not extend to the moral or personal ownership of authorship, which remains with the creator. Spanish law, therefore, maintains a strict distinction between authorship and economic ownership, allowing the transfer of patrimonial rights while preserving authorship as an inalienable personal attribute.

The United States, by contrast, recognizes no presumption in favor of the claimant. The evidentiary burden rests entirely on the applicant, who must demonstrate a human creative contribution as a prerequisite for registration. Failure to provide such proof results in the denial of protection. This position has been consistently reaffirmed in the official policies of the U.S. Copyright Office and through recent administrative and judicial decisions, establishing a regime of strict evidentiary scrutiny for works containing AI-generated material. The absence of sufficient human intervention unambiguously results in exclusion from the copyright registry (U.S. Copyright Office 2023b, 2025).

The United Kingdom also presents a sui generis mechanism for computer-generated works. Section 9(3) of the Copyright, Designs and Patents Act 1988 stipulates that, in cases where no identifiable human author exists, authorship shall vest in the person who made the necessary arrangements for the creation of the work. The provision does not presume human authorship but rather focuses on functional responsibility. The standard is instrumental rather than subjective: the system operator is recognized as the rights holder if able to demonstrate control over the overall creative process, even without direct intervention in every aspect of the outcome. This model omits any requirement of artistic intention or personal expression, a characteristic that markedly differentiates it from continental frameworks.

The comparative landscape, therefore, reveals the absence of a single, unified method for proving authorship in AI-assisted creation. Each legal system assesses human intervention through its own normative lens, leading to a plurality of evidentiary standards. Some jurisdictions emphasize the subjective connection between the work and its creator, whereas others prioritize the demonstrable traceability of the creative process. Across all frameworks, however, a common principle persists: authorship is not automatically presumed or conferred in contexts of advanced automation. Individuals seeking protection must be prepared to substantiate their creative role, demonstrate originality, and prove that, beyond the use of the tool, they exercised intellectual control over the process. Such a requirement should not be perceived as a barrier but rather as a safeguard against the opacity of generative technologies and as a means—perhaps the only viable one at present—to preserve the centrality of the human agent within artistic and literary creation.

5.2. Emerging Convergences in Administrative and Judicial Practice

Despite the theoretical and normative divergences that persist between copyright regimes of continental origin and those grounded in the Anglo-Saxon legal tradition, comparative analysis reveals an emerging convergence toward a shared minimum standard in the protection of works generated with the assistance of artificial intelligence. The alignment does not stem from formal legislative harmonization or supranational mandates, but rather from a gradual and largely pragmatic evolution of administrative practices, doctrinal approaches, and judicial reasoning. What only a few years ago appeared to be a zone of regulatory ambiguity is now consolidating into a field governed by principles that, although articulated differently, reflect a growing transnational consensus.

A central tenet of this convergence is the requirement of substantial human participation as an indispensable condition for the recognition of authorship. The principle, far from novel, has been explicitly reaffirmed by both the United States Copyright Office and the European Union Intellectual Property Office. The latter emphasizes in its technical report that artificial intelligence should be regarded as a tool serving the author, not as a substitute capable of autonomous creative authorship (EUIPO 2025). Legal protection is therefore contingent upon demonstrating that the human contributor has exercised verifiable creative control over the result. A parallel position is expressed in the administrative guidance issued by the U.S. Copyright Office, which requires that applicants clearly identify the elements of a work produced automatically by artificial intelligence and those that can be attributed to human creative input, excluding from registration any component generated without substantial human intervention (U.S. Copyright Office 2023a).

The coherence of this position is further evidenced in concrete cases such as Zarya of the Dawn, where protection was granted only to those components that embodied human creative decisions, including the narrative text and the overall arrangement of content. The images generated using the Midjourney system were excluded on the basis that no verifiable connection could be established between the final visual output and an intentional, conscious act of human creativity (U.S. Copyright Office 2023b). Similarly, in Thaler v. USPTO (U.S. Court of Appeals for the Federal Circuit 2022), it was confirmed that works produced autonomously by artificial intelligence lack the requisite human authorship to qualify for protection. Across both contexts, a unifying criterion emerges: the traceability of human decision-making throughout the creative process constitutes the fundamental element distinguishing mere algorithmic execution from legally recognized creation.

The same convergence can be observed in the identification of shared parameters used to evaluate originality, even when the philosophical foundations of the respective legal systems diverge. Within the European Union, originality is conceived as the expression of an author’s individuality, requiring that the work embody the author’s personal imprint and constitute an autonomous intellectual creation, as recognized in Infopaq International A/S v. Danske Dagblades Forening (CJEU 2009) and Eva-Maria Painer v. Standard Verlags GmbH (CJEU 2011). German doctrine articulates an analogous standard through the notion of Schöpfungshöhe, which demands a non-trivial level of individual creativity as a threshold for protection (§2(2) Urhebergesetz). In the United States, where the legal system adopts a more utilitarian perspective, the decision in Feist Publications, Inc. v. Rural Telephone Service Co. (U.S. Supreme Court 1991) similarly established that copyright requires “some degree of creativity,” rejecting the notion that mere labor or factual compilation suffices to confer protection.

Similarly, the United Kingdom, through its jurisprudence and the regulatory framework established by the Copyright, Designs and Patents Act of 1988, applies a logic that requires the exercise of skill, labor, and judgment in the creative process. Although such a standard does not demand that a work reflect the author’s personality, it presupposes the existence of a discernible intellectual intervention that cannot be ascribed to a purely automatic mechanism. This criterion has allowed courts, as demonstrated in Nova Productions v. Mazooma Games (Court of Appeal 2007), to deny protection where the user’s engagement with the system was merely mechanical rather than creative.

In administrative and hybrid systems, such as those embodied in the practice of the European Union Intellectual Property Office and in the United Kingdom’s specific framework for computer-generated works under section 9(3) of the Copyright, Designs and Patents Act, a more flexible approach to originality thresholds has gradually emerged. However, even under these models, a consistent boundary persists: outputs generated without sufficient human intervention do not receive automatic protection. A work may originate from an intricate technical or algorithmic process, but its protectability depends on whether the human contribution can be identified, traced, and substantiated. Flexibility, therefore, does not entail a relaxation of authorship requirements, but rather an acknowledgment of the plurality of ways in which creativity manifests in technology-mediated contexts.

The convergence of legal reasoning across jurisdictions should not be understood as the erasure of doctrinal distinctions. The European Union continues to uphold a conception of authorship rooted in the dignity of the creator and the intrinsic value of individual expression. The United States, by contrast, adheres to a utilitarian orientation that emphasizes the incentive to create and the broader social value of cultural production. Yet both systems converge on a fundamental premise: legal protection must be reserved for cases where human intervention plays an effective role, excluding results that can be autonomously attributed to the technical operations of artificial intelligence.

The point of convergence between these distinct paradigms may be found in the notion of creative responsibility. Although articulated differently across legal traditions, each framework recognizes that copyright ultimately serves to attribute legal and moral consequences to a human act of creation. Artificial intelligence, regardless of its technical sophistication, does not possess intentionality, judgment, or aesthetic discernment. Consequently, unless a profound transformation occurs in the ontological and normative understanding of authorship, the human creator will remain the irreplaceable axis of the system. Such an affirmation should not be construed as a rejection of the creative potential inherent in generative technologies, but rather as a reminder that their legal significance, as with any tool, depends entirely on the intellectual agency of those who employ them.

In an environment where the boundaries between creation and automation are increasingly permeable, the emergent consensus regarding the necessity of meaningful human intervention functions as a normative anchor. Even though legal systems may diverge in defining the precise contours of originality, there is growing agreement that authorship cannot exist without human agency. Such recognition, while not resolving the full complexity of the challenges posed by artificial intelligence, establishes a foundational principle from which legal reflection can proceed: the protection of human participation in creative processes is not a mechanism of technological exclusion, but a condition that gives conceptual and juridical meaning to the very idea of the work.

6. Implications for the Fundamental Concepts of Authorship and Originality

The conceptual evolution of creation in the contemporary context, marked by the emergence of generative artificial intelligence systems, has transformed not only the technical and operational dimensions of the creative process but also the very foundations of authorship attribution. Legal frameworks that traditionally defined the author as a direct and personal embodiment of individual expression, particularly within systems of continental origin, are now being challenged by models of cultural production in which the human role may be indirect, mediated, or even marginal.

Continental law has historically conceived authorship as a manifestation of the creator’s personality, establishing a subjective and non-transferable bond between the author and the work. This conception finds its clearest normative expression in Article L121-1 of the French Code de la propriété intellectuelle, which recognizes inalienable moral rights precisely because it considers the work a projection of the author’s individuality. The same rationale has been endorsed by the Court of Justice of the European Union, which, in Eva-Maria Painer v. Standard Verlags GmbH (CJEU 2011), affirmed that a work is original when it reflects the author’s free and creative choices, that is, when a “personal imprint” is perceptible in the final result.

A gradual shift has nevertheless occurred from this expressive model toward a more functional logic centered on indirect creative control. The decision in Zarya of the Dawn (U.S. Copyright Office 2023b) exemplifies this transformation, since protection was not granted on the basis of direct subjective expression but rather on the author’s capacity to organize, structure, and select content generated through AI tools. The creative acts identified in the case, such as the arrangement and curation of materials, were deemed sufficient to justify authorship, even though the technical execution was mediated by automation. This development suggests a reformulation of the author–work relationship, whereby technological mediation does not preclude protection as long as active human direction can be verified throughout the process.

The relationship between creator, tool, and product has thus been reconfigured in fundamental ways. While in classical theory the tool—be it a brush, a camera, or software—was understood as entirely subordinate to the author’s will, contemporary AI systems operate with a degree of autonomy that introduces new complexities for the attribution of authorship. Under British law, section 9(3) of the Copyright, Designs and Patents Act of 1988 allows for the possibility that a computer-generated work may lack an identifiable human author, assigning authorship instead to the individual who made “the necessary arrangements” for its creation. This pragmatic approach contrasts with the more restrictive positions of continental Europe and the United States, where significant and verifiable human intervention remains indispensable, and the mere activation or generic use of AI tools is insufficient to constitute a protectable act of creation (U.S. Copyright Office 2023a; EUIPO 2025).

The emergence of artificial intelligence has also generated a form of creative abundance that destabilizes the traditional notion of originality. The ease with which AI systems produce vast quantities of formally novel but structurally derivative content has diminished the centrality of “novelty of result” as a defining criterion. Jurisprudence in both Feist Publications, Inc. v. Rural Telephone Service Co. (U.S. Supreme Court 1991) and Infopaq International A/S v. Danske Dagblades Forening (CJEU 2009) clarified that originality does not require absolute novelty, but a minimal level of human creativity. This principle acquires renewed relevance in AI-assisted contexts, where the primary value lies not in the reproducible aesthetic output but in the human intellectual process guiding it. A qualitative notion of originality, emphasizing human intervention over the mere external characteristics of the final product, is therefore gaining prominence.

A central difficulty arises in distinguishing between “algorithmic originality,” which results from statistical recombination and learned patterns, and “human originality,” which derives from deliberate aesthetic and intellectual choices. German law addresses this distinction through the Schöpfungshöhe principle, codified in §2(2) of the Urhebergesetz, which requires a sufficient degree of personal creativity for protection, as confirmed in the Inkasso-Programm decision (BGH 1985). In the United States, Thaler v. USPTO (U.S. Court of Appeals for the Federal Circuit 2022) reaffirmed that a work generated entirely by AI without human intervention cannot be protected under copyright. Consequently, the differentiation between human and automated production has become a foundational condition for contemporary legal analysis.

New creative roles have emerged within this evolving framework. The figure of the prompt engineer exemplifies a technical and artistic mediator between the AI system and the final product. The legal relevance of such intervention ultimately depends on the degree of intentionality, expertise, and judgment demonstrated in the process. British jurisprudence, through the standard of skill, labor, and judgment applied in Nova Productions v. Mazooma Games (Court of Appeal 2007), provides a basis for recognizing this form of assisted creativity, provided it exceeds the level of mere mechanical execution. The design of complex and semantically meaningful prompts capable of decisively influencing the outcome can, therefore, under certain conditions, be regarded as an act of authorship.

A similar interpretative challenge emerges in the case of curators or managers of generative processes, whose creative contribution lies in the selection, combination, and adaptation of AI-generated outputs. When such curation involves discernible creative decisions, as acknowledged in Zarya of the Dawn, it may constitute authorship founded on meaningful organization rather than material production. The decisive factor remains the exercise of intellectual control over the process, rather than passive facilitation.

Gradual models of authorship are consequently beginning to take shape. Where an individual exerts dominant direction—by defining parameters, iteratively selecting outputs, and substantially editing the results—both U.S. and European doctrine recognize the possibility of full authorship. Where human intervention is relevant but partial, frameworks may allow for shared or fragmented attribution, as illustrated by the partial protection granted to the textual and structural components in Zarya of the Dawn. Conversely, when human participation is minimal, generic, or merely instrumental, protection is denied. Both Thaler v. USPTO and the European requirements of a “personal imprint” reaffirm that without demonstrable creative control, authorship cannot exist.

Evaluation criteria have likewise evolved. Traditional analysis focused on the final work as a static object, valuing its form and apparent novelty. Contemporary reasoning increasingly emphasizes the creative process itself. The U.S. Copyright Office, in the Zarya decision, recognized not the visual result but the author’s creative trajectory leading to the textual composition. Likewise, the CJEU in Painer and Infopaq evaluated originality through the cumulative set of creative choices made by the author. This reasoning acknowledges that in contexts where machines participate actively, authorship cannot reside exclusively in the product but must be found in the human direction and intellectual shaping of that product.

Intentionality, therefore, emerges as a constitutive element of creativity. The deliberate selection of inputs, conscious supervision of the generative process, and critical curation of results form the evidentiary basis for attributing authorship. The guidelines issued by both the U.S. Copyright Office and the EUIPO converge on this interpretation, emphasizing that legal evaluation must consider human intention and identifiable creative input rather than the aesthetic quality of the outcome. Such an approach opens the way for a more contextual and equitable understanding of protection, capable of recognizing the diverse modalities through which creativity is now exercised in the age of generative technologies.

7. Toward a Harmonized and Human-Centered Framework for AI-Generated Creativity in International Copyright Law

7.1. Compatibility of AI-Created Works with the International Copyright Architecture

Within the international intellectual property system, generative artificial intelligence has emerged as a phenomenon that challenges not the foundational principles of copyright, but their concrete application to new creative contexts. The Berne Convention, the cornerstone of the multilateral framework, establishes that a protected work must result from an “intellectual creation,” a concept that, although undefined in detail, has been almost universally interpreted as the manifestation of conscious and individual human activity. The notion of authorship implicit in Article 2 of the Convention, reinforced by the general structure of the system and by the moral rights enshrined in Article 6bis, which are attributable exclusively to natural persons, excludes by definition the possibility of recognizing non-human entities, including artificial intelligences, as authors. Intellectual creation, under this logic, presupposes volition, aesthetic or communicative intention, and the capacity for autonomous choice, all attributes that remain uniquely human.

However, such a conception does not prevent the States party to the Convention from developing specific approaches to works that are assisted, though not entirely generated, by artificial intelligence. The margin of discretion, while limited, remains significant. On one hand, the Convention imposes no formalistic conditions regarding the means of creation, which allows for a flexible interpretation of what may qualify as a “work.” On the other hand, it does not provide a closed definition of originality or an exhaustive taxonomy of protectable creations, thereby permitting domestic systems to establish additional requirements for human intervention without contravening the treaty. Moreover, since protection under the Convention is governed by the principle of nationality rather than material reciprocity, States retain the right to exclude outputs generated exclusively by artificial intelligence from protection, provided that such exclusions do not amount to arbitrary discrimination between nationals and foreigners.

A comparable interpretative flexibility can be observed within the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) adopted under the World Trade Organization. This instrument establishes a minimum standard of protection derived from the Berne Convention, except with respect to moral rights, while granting member States broad regulatory autonomy to define national conditions, as long as they comply with the principles of national treatment and non-discrimination. The TRIPS framework, therefore, permits the introduction of stricter originality criteria or the creation of sui generis protection regimes, provided that these do not undermine the fundamental rights of human authors or constitute unjustified barriers to trade.

Consequently, the current absence of specific international rules governing AI-generated or AI-assisted works should be understood not as a systemic deficiency, but as an opportunity for regulatory innovation. In domains not yet consolidated by multilateral consensus, States enjoy strategic freedom to design their own legal models. This flexibility is particularly valuable in jurisdictions where legislative processes are slow to adapt, yet where technological transformation demands timely regulatory responses. The adoption of experimental mechanisms, such as tiered protection systems based on degrees of human intervention, would not only be consistent with international obligations but could also position certain jurisdictions as normative leaders in developing adaptive, inclusive, and forward-looking copyright regimes.

In that sense, one of the central strategic choices for legislators and policymakers, especially in developing countries, lies in deciding between a teleological reinterpretation of existing norms and comprehensive legislative reform. The former, more immediate and less disruptive approach, involves applying existing principles to novel technological realities through an evolutionary reading of expressions such as “intellectual creation” or “original work.” This interpretative flexibility has been reflected in the Spanish Supreme Court’s (2007) Judgment 696/2007, which reaffirmed that authorship corresponds to the natural person who creates the work while acknowledging that the ownership of economic rights may belong to an employer by virtue of legal assignment, thereby recognizing a functional differentiation between creative and proprietary roles.

The latter alternative, more complex yet potentially more stable, entails legislative reform that explicitly incorporates AI-mediated creation into statutory law. Such reform would involve defining the role of artificial intelligence as a creative instrument, setting minimum thresholds for human involvement, and differentiating legal consequences according to the level of human participation. This approach could enhance legal certainty for creators, institutions, and administrative authorities, though it would require the difficult convergence of technical expertise, political will, and cultural consensus.

A particularly effective interim strategy may consist of developing harmonized administrative guidelines at the regional level. Such soft-law instruments have proven successful in comparable contexts, as shown by the EUIPO guidelines issued between 2023 and 2025, which stipulate that AI-assisted works are protectable only when substantial human intervention can be identified, documented, and evaluated (EUIPO 2025). Similarly, the U.S. Copyright Office’s guidance of March 2023 requires applicants to disclose the use of artificial intelligence in the creative process, identify automatically generated elements, and provide evidence of human creative control over the final product (U.S. Copyright Office 2023a). Although lacking binding legal authority, these administrative frameworks facilitate case-by-case assessments, foster institutional consistency, and operate as transitional instruments pending legislative reform.

7.2. Substantial Creative Direction as a Harmonizing Principle in International Copyright Law

The development of a stable and adaptive legal framework for the protection of works generated or assisted by artificial intelligence demands a careful equilibrium between the normative foundations of copyright and the structural transformations of technological creativity. The core of this challenge lies in identifying an authorship standard capable of distinguishing authentic human expression from algorithmic generation. A criterion based on substantial creative direction provides a robust foundation for such differentiation, as it encapsulates verifiable human control over the generative process, demonstrable creative input, and a discernible expressive intentionality. Through this approach, the essence of copyright protection, grounded in intellectual agency and moral accountability, remains intact within hybrid human–machine environments.

Comparative legal doctrine reinforces the validity of this approach. The Court of Justice of the European Union has repeatedly affirmed that protection under copyright arises only where an author’s own intellectual creation is evident, reflected in free and creative choices that bear a personal imprint and convey individuality (CJEU 2009, Infopaq C-5/08; CJEU 2011, Painer C-145/10). The principle thereby consolidates an anthropocentric conception of authorship in which human creativity remains the decisive factor in defining originality. In harmony with this jurisprudence, the analysis of intellectual creation as a normative safeguard sustains the indispensable link between human intention and legal authorship, preventing the attribution of rights to autonomous generative systems (Gaffar and Albarashdi 2025). Similarly, the emphasis on the necessity of an identifiable expressive contribution underscores that copyright must remain anchored in human intentionality and moral responsibility, as detaching protection from the human source would dissolve the ethical and conceptual coherence of authorship (Gervais 2021).

Further depth is provided by arguments that interpret the European copyright system as already equipped to accommodate AI-assisted creativity. The analysis of human participation within multiple creative phases—conception, execution, and redaction—demonstrates that originality is not negated by technological mediation, provided that human agency determines the expressive outcome (Hugenholtz and Quintais 2021). This reasoning affirms that creative authorship remains discernible when an individual exercises discretion in directing or curating AI outputs, thus aligning with the broader framework of substantial creative direction that privileges human intentionality as the threshold of legal protection.

In contrast, a different perspective interprets copyright as a fluid mechanism for achieving social equilibrium rather than a strict system of human authorship. The functional conception of fair use as a tool for balancing protection and access underscores the adaptability of copyright to new technological realities (Burrell and Coleman 2005). Likewise, the characterization of fair use as a flexible legal principle that advances innovation and democratic participation within digital ecosystems highlights its capacity to harmonize proprietary control with social utility (Elkin-Koren and Netanel 2021). Yet, the expansion of such flexibility to encompass AI-generated content risks undermining the ontological basis of copyright, as the absence of human intentionality would transform copyright from a system rewarding creativity into a mechanism protecting automated processes. The overextension of fair use beyond its social corrective function could thereby erode the conceptual integrity of authorship and dilute the distinction between creative and computational production.

Maintaining an anthropocentric threshold ensures that legal protection corresponds to genuine human effort and expressiveness. The European notion of originality, centered on the author’s intellectual creation, embodies this principle by requiring traceable human freedom and decision-making within the expressive form of the work. The collapse of this nexus, warned against in theoretical analysis, would dismantle the justificatory structure of copyright and compromise its ethical legitimacy (Gervais 2021; Gaffar and Albarashdi 2025). Accordingly, the delineation of authorship must remain contingent upon the capacity to identify the human creative act within the generative process, not merely its technological execution.

From this perspective, a graduated protection model emerges as a rational and proportionate framework. Full copyright should apply where human participation is dominant and demonstrably shapes the expressive result, as when the creator configures parameters, selects datasets, and edits AI-generated material to achieve a distinct artistic vision. In cases of significant yet insufficient human involvement, a limited sui generis regime may be appropriate, following the model of section 9(3) of the United Kingdom’s Copyright, Designs and Patents Act of 1988, which attributes authorship to the person making the necessary arrangements for creation. By contrast, purely automated productions resulting from mechanical or instrumental human input should remain in the public domain, in accordance with recent jurisprudence denying protection to autonomous AI outputs (Thaler v. USPTO, U.S. Court of Appeals for the Federal Circuit 2022).

Acknowledging the public domain as the natural repository of non-human creations strengthens, rather than restricts, innovation, since it preserves the balance between private incentive and collective access. The absence of human authorship must not extend the scope of copyright but reaffirm its raison d’être as the legal recognition of intellectual and moral creativity. The criterion of substantial creative direction, coupled with a tiered protective structure, thus offers a coherent and ethically consistent foundation for contemporary authorship. It integrates technological evolution into the normative architecture of copyright, ensuring that the law continues to reflect the enduring centrality of human intention, expressive agency, and accountability within the creative order of the algorithmic age.

7.3. Procedural Harmonization and Institutional Adaptation

The integration of artificial intelligence into creative processes has introduced significant complexity into existing intellectual property frameworks, particularly concerning the determination of authorship and originality. This transformation demands an urgent adaptation not only of substantive legal norms but also of the administrative procedures and evidentiary mechanisms employed by national intellectual property offices. For such adaptation to be both legitimate and effective, it must be grounded in a model that combines transparency obligations with evidentiary standards tailored to the cultural and technological characteristics of each jurisdiction, alongside a progressive institutional professionalization.

An initial procedural measure should be the implementation of a mandatory disclosure system requiring applicants to declare the use of artificial intelligence tools during the creative process. At the time of registration, applicants should be required to specify whether AI systems were employed, identify the portions of the content produced through such tools, and detail the degree of human intervention in each segment. This approach draws support from the current guidelines of the U.S. Copyright Office (2023a) and the recommendations issued by the European Union Intellectual Property Office (EUIPO 2025). Both institutions converge on the principle that the use of artificial intelligence does not, in itself, exclude protection, but necessitates a detailed examination of human contribution to determine eligibility.

Nevertheless, disclosure alone is insufficient. The system must also require the submission of verifiable documentation of the creative process. Applicants should present tangible evidence such as the prompts used to generate content, records of iterative refinements made through the AI tool, and reasoned justifications for the human creative decisions that guided the process. The objective of this requirement is not to increase bureaucratic burdens but to establish objective criteria that enable examiners to more accurately assess the degree of human control and authorship. The experience of Zarya of the Dawn illustrates the practical utility of this approach, since protection was granted only to those portions of the work whose configuration could be traced to identifiable and verifiable human choices (U.S. Copyright Office 2023b).

From an evidentiary standpoint, not all works present the same challenges of attribution, which justifies differentiated standards of proof depending on the nature of the creation. For visual works, such as photographs, illustrations, or digital graphics, the analysis should focus on elements including the selection of source materials, compositional design, manual adjustments, and aesthetic refinements. In contrast, for literary or textual works, assessment should emphasize narrative structure, stylistic coherence, and the semantic configuration of the discourse, dimensions that reveal the author’s creative imprint. The concept of a “personal footprint,” recognized by the Court of Justice of the European Union in Painer (CJEU 2011, C-145/10), provides an instructive framework for such assessments, as it allows originality to be understood not as an absolute quality but as the concrete expression of individual creativity through discernible aesthetic decisions.

Any reform of this nature presupposes, however, a corresponding enhancement of institutional capacity within intellectual property offices. It is imperative to invest in specialized training for examiners responsible for assessing applications involving AI-generated components. Such training should encompass comparative analysis of international standards, methodological tools to distinguish between automated and human-directed content, and familiarity with the technical parameters underlying generative models. Expertise in the evaluation of creative control, as well as proficiency in emerging digital technologies, constitutes a necessary safeguard against arbitrary or inconsistent administrative decisions. The establishment of a coherent, technically informed evaluative framework is indispensable to ensuring both procedural fairness and normative consistency across cases.

Ultimately, the challenge extends beyond the reform of substantive law. It requires the translation of normative principles into fair, transparent, and context-sensitive administrative practice. Transparency in the disclosure of AI use, systematic documentation of the creative process, differentiated evidentiary standards, and professional specialization within intellectual property offices represent the foundational pillars of a system designed to uphold human creativity while acknowledging the growing role of the artificial in contemporary authorship. Addressing these challenges through structured institutional reform would not only enhance the credibility and coherence of copyright enforcement but also reaffirm the human dimension of creativity as the cornerstone of intellectual property in the age of artificial intelligence.

8. Conclusions

The technological evolution represented by generative artificial intelligence systems has profoundly disrupted the doctrinal foundations of copyright, compelling a reconceptualization of authorship and originality. Evidence from both continental and Anglo-American traditions confirms the persistence of an anthropocentric conception that inextricably links authorial protection to the existence of significant human intervention. Such a link is not incidental but reflects the essentially human nature underlying any legally protectable creative act, even in contexts characterized by intense technological mediation.

The principal theoretical contribution of the analysis lies in the gradual and contextual characterization of human intervention within AI-assisted creative processes. Authorship emerges not as a binary category but as a continuum that extends from mere instrumental activation to substantive creative control. The formulation of the standard of substantial creative direction, understood as the verifiable combination of effective control, creative input, and expressive intentionality, provides a coherent conceptual framework that transcends formal divergences between legal traditions. This approach enables a qualitative assessment more consistent with the realities of contemporary creation mediated by intelligent systems.

From a pragmatic perspective, the operationalization of the proposed model requires procedural adaptation within registration and administrative systems. The establishment of a mandatory disclosure regime for the use of AI tools, accompanied by documentation requirements calibrated to the type of work and the degree of human intervention, constitutes an essential step toward ensuring legal certainty for creators employing such technologies. Likewise, the tiered protection model, which recognizes full, limited, or no rights depending on the level of creative control exercised, offers a differentiated system of protection that harmonizes the promotion of innovation with the preservation of the intrinsic value of human creativity.

It must be acknowledged, however, that the normative proposal advanced here encounters inherent limitations linked to its legal-comparative perspective. The analysis has concentrated primarily on the formal aspects of authorship attribution, without sufficiently addressing the distributive and structural dimensions that shape the technological field. Asymmetries in access to advanced AI tools, corporate concentration in algorithmic development, and the geopolitical configurations conditioning technology transfer constitute critical variables that, while extending beyond the regulatory domain, significantly affect the practical effectiveness of any protective framework.

Future research must therefore transcend legal formalism by incorporating multidisciplinary perspectives capable of interrogating the intersections among authorship, automation, and creative autonomy. The exploration of the implications of AI for cultural diversity, technological sovereignty, and the sustainability of local creative ecosystems emerges as an urgent agenda. Equally necessary is the development of evaluative methodologies capable of discerning, with greater technical precision, the actual magnitude of human intervention in increasingly complex works, where the boundaries between creation and generation become progressively blurred. Empirical assessment of the effects of different regulatory models on creative practices likewise represents an indispensable horizon for future inquiry.

At a more reflective level, the enduring tension between the humanist tradition of copyright and the advent of algorithmic creativity raises questions that reach beyond the legal or technical to the anthropological. The defense of a conception of authorship grounded in human agency should not be misinterpreted as nostalgic resistance to technological progress but rather as an affirmation that legally protectable originality continues to reside not in mere combinatorial novelty but in the distinctively human capacity to endow creation with meaning, intentionality, and expressiveness. In the age of algorithmic reproducibility, the unique value of human judgment, sensibility, and intentional purpose acquires renewed significance. The challenge for contemporary legal systems, particularly in emerging contexts, lies in preserving this humanistic core of authorial protection while critically and reflectively integrating new technologies into the evolving landscape of creativity.