Click here to read our latest publication "AI, Fanart and Intellectual Property – an Analysis of Ownership in The Age of Machine Creativity" by Martine Mussies
Home > Publications > "AI, Fanart and Intellectual Property – an Analysis of Ownership in The Age of Machine Creativity"
May 26th 2025
AI, Fanart and Intellectual Property – an Analysis of Ownership in The Age of Machine Creativity
By Martine Mussies
Martine Mussies is an artistic researcher and autistic academic based in Utrecht, the Netherlands. She is a PhD candidate at the Centre for Gender and Diversity at Maastricht University, where she is writing her dissertation on The Cyborg Mermaid. Martine is also part of SCANNER, a research consortium aimed at closing the knowledge gap on sex differences in autistic traits. In her #KingAlfred project, she explores the online afterlives of King Alfred the Great, and she is currently working to establish a Centre for Asia Studies in her hometown of Utrecht. Beyond academia, Martine is a musician, budoka, and visual artist. Her interdisciplinary interests include Asia Studies, autism, cyborgs, fan art and fanfiction, gaming, medievalisms, mermaids, music(ology), neuropsychology, karate, King Alfred, and science fiction. More at: www.martinemussies.nl and LinkedIn.
Introduction: Rewriting Authorship in the Digital Continuum: AI-Generated Fanart and the Evolution of Copyright Frameworks
In an age where artificial intelligence has evolved from distant abstraction to everyday creative collaborator, the legal scaffolding surrounding authorship, ownership, and derivative works demands critical reevaluation. While AI represents the latest development in a long continuum of creative technologies that challenge traditional notions of authorship, generative models like DALL-E, Midjourney, and Stable Diffusion have intensified familiar tensions within copyright law to unprecedented degrees. At the vanguard of this transformation lies AI-generated fanart: digital artworks derived from popular culture (video games, cinematic universes, retrocomputing) conjured through algorithmic processes initiated by fans, often with minimal direct artistic input.
This paper builds upon my earlier research into the cultural significance of AI-enabled fan practices (Mussies, 2023), but shifts the analytical lens towards the legal domain—specifically, towards copyright law's capacity to accommodate non-human or hybrid authorship. It adopts an integrated theoretical framework that synthesizes three complementary approaches: doctrinal legal analysis, cultural studies perspectives on participatory creation, and philosophical inquiries into distributed authorship. Rather than treating these as separate lenses, this integration reveals a more comprehensive understanding of AI-generated fanart's position within established legal and cultural structures.
Three interrelated questions form the spine of this inquiry: How do existing doctrines of authorship and originality accommodate technologically-mediated creative processes? Does prompt engineering constitute meaningful authorship or mere facilitation? And how can we situate AI-generated fanart within the historical development of fair use and derivative work doctrines? To address these, this paper adopts a methodologically rigorous approach, anchored in case law, statutory analysis, and regulatory guidance, enriched by comparative insights from the United Kingdom, the European Union, the United States, and Japan.
The significance of this inquiry lies not merely in proclaiming a revolutionary break with past understandings, but in carefully tracing how technological change tests the elasticity of existing legal frameworks. Fanart has long occupied a liminal legal space—what scholars term a "tolerated use" zone—celebrated as homage yet precariously poised on the edge of infringement. The introduction of generative AI has deepened this ambiguity. Platforms now allow users to generate stylized portraits of established characters or imagined artifacts in specific aesthetic styles, often indistinguishable from traditional digital art. These creations circulate widely in online communities, where they are celebrated, modified, and shared—yet increasingly attract takedown notices and legal scrutiny from IP holders.
What distinguishes this analysis is its attention to both continuity and rupture. While positioning AI generation on a continuum with other forms of technologically-assisted creation, it acknowledges the unique challenges posed by systems that can generate content with minimal human direction. Human-machine collaboration emerges as a key site of contestation, where questions of agency, intent, and creative contribution become increasingly complex. While some proponents advocate for expanding copyright's definition of authorship to include AI outputs, others warn that such a move may dilute human creativity or exacerbate exploitation by corporate actors. AI, after all, is not inherently good or bad—it is a tool whose ethical valence depends on its application: it can be harnessed for benevolent purposes, such as an intelligent data platform to resolve global food shortages (Capgemini, 2019), or for harm, such as stalking and terrorising others online (Mussies, 2025).
In mapping this terrain, the paper proceeds in five parts. First, it establishes the historical and legal context of fanart, delineating the conceptual stakes and distinguishing between human-made, digitally assisted, and fully generative works. Second, it analyzes specific legal conflicts involving AI-generated fanart, situating them within established doctrinal traditions. Third, it provides a comparative analysis of how different jurisdictions apply existing principles to questions of machine-assisted creation. Fourth, it examines philosophical perspectives on distributed authorship and the ethical implications of assigning legal personhood to non-human agents. Finally, it concludes with targeted recommendations for doctrinal refinements that respect both creative innovation and the moral rights of original creators in an AI-mediated cultural economy.
This is not merely a legal conundrum, but a cultural and philosophical one. By examining AI-generated fanart as a critical testing ground for the evolving contours of intellectual property—where law, culture, and computation intersect in ways both exhilarating and destabilizing—this paper contributes to a more nuanced understanding of authorship in the digital age.
2. Fanart in Legal and Cultural Perspective
2.1 Historical Context: Fanart Within Copyright's Evolutionary Development
Fanart has long occupied a liminal position in the landscape of copyright law: neither entirely legal, nor inherently illicit. Since the conceptual origins of copyright in the Statute of Anne (1710), legal frameworks have continuously adapted to new technologies and creative forms—from photography and recorded music to software and digital networks. Within this evolutionary continuum, fanart represents not an anomalous legal problem but a particularly illustrative case of copyright's ongoing negotiation with changing creative practices.
Often created as a form of homage, parody, or intertextual tribute, fanart involves the unauthorized visual reinterpretation of characters, settings, or styles drawn from copyrighted works. In most jurisdictions, such creative acts technically constitute derivative works, and thus fall within the scope of exclusive rights reserved for the original copyright holder. However, enforcement has historically been inconsistent, shaped as much by pragmatic non-intervention as by formal legal doctrine.
Fanart occupies a position within what scholars call copyright's "gray economy" of tolerated uses—creative practices that persist through implicit forbearance rather than explicit authorization. This legal liminality is not unique to fanart but characterizes numerous creative practices throughout copyright history. Jazz musicians' improvisations on copyrighted standards, literary adaptations before licensing became standardized, and artistic appropriation movements all inhabited similar zones of contested legitimacy before courts and customs clarified their status.
The legal ambiguity stems in part from the inherently expressive and often non-commercial nature of fanart. In jurisdictions like the United States, the fair use doctrine (codified in the 1976 Copyright Act but developed through earlier case law) provides a flexible framework for assessing transformative secondary uses. The four-factor test—examining purpose, nature of the original work, amount used, and market effect—was not designed specifically for fanart but has proven adaptable to various forms of cultural reinterpretation. In the United Kingdom and the EU, similar protections may arise under the parody or pastiche exceptions, though these are narrowly construed and often contested.
These doctrines have evolved incrementally through judicial interpretation rather than revolutionary reconception, demonstrating copyright's capacity for adaptation within established paradigms. At the same time, fanart exists in a quasi-tacit economy of permission, in which rights holders tolerate—and occasionally even encourage—its production, recognizing its promotional or community-building potential.
2.2 Cultural Legitimacy and Legal Recognition: Convergent or Parallel Systems?
The apparent disconnect between fandom's cultural norms and copyright's doctrinal structures reflects a broader pattern in which legal and social understandings of creativity exist in productive tension. While law regards unauthorized derivation with suspicion, fandom communities view such acts through an entirely different lens—one grounded in affect, identity, and participatory engagement. In these spaces, creating fanart is not a legal violation, but a form of cultural citizenship: a way of contributing to, negotiating with, and sometimes resisting the "canon" defined by corporate producers.
Henry Jenkins's theory of convergence culture provides a useful framework not because it supersedes legal analysis, but because it helps explain how social practices often anticipate doctrinal evolution. When Jenkins describes fans "poaching" from corporate media texts and repurposing them for their own interpretive communities, he identifies a dynamic that has historical parallels in numerous contexts where cultural practices eventually influenced legal interpretation. This process often includes gender-swapped depictions, queer reinterpretations, or recontextualization across genres—practices that, while legally vulnerable, are culturally valid and socially generative.
Rather than positioning fandom's participatory ethic as fundamentally opposed to copyright's exclusivity principles, we can understand them as complementary systems that gradually influence each other. Fan practices that emphasize transformation, attribution, and non-commercial sharing have parallels in copyright's emerging recognition of transformative use, moral rights, and non-commercial exceptions. This convergence suggests not a paradigm shift but a familiar pattern of mutual adaptation.
Moreover, as Jessica Hautsch (2022) and others have observed, fanart constitutes a form of cognitive engagement with the source material—an active process of re-imagining rather than passive consumption. This observation aligns with copyright's long standing distinction between protected expression and unprotected ideas. When fans reinterpret characters or settings, they engage primarily with conceptual elements rather than specific expressive choices—a distinction that copyright doctrine has recognized since Baker v. Selden (1879) established the idea/expression dichotomy. This helps explain why many fan creators feel a moral, if not legal, claim to legitimacy. The fan creator is not stealing; they are responding—often in ways that enrich the narrative ecosystem of the source work.
2.3 Technological Evolution and Legal Adaptation: Historical Precedents and Conflicts
Despite its cultural legitimacy, fanart has not been immune to legal conflict. The introduction of new creative technologies has consistently prompted reconsideration of copyright's boundaries without necessitating its conceptual reinvention. Before the advent of generative AI, high-profile cases and platform controversies illustrated the inherent instability of fanart's legal position.
Photography's emergence in the 19th century raised questions about authorship in mechanically-produced images that parallel contemporary debates about AI generation. In Burrow-Giles Lithographic Co. v. Sarony (1884), the U.S. Supreme Court recognized photographic works as protectable by emphasizing the photographer's creative choices in "selecting and arranging the costume, draperies, and other various accessories... arranging the subject so as to present graceful outlines, arranging and disposing the light and shade, [and] suggesting and evoking the desired expression." This emphasis on human selection and arrangement provides a doctrinal precedent for assessing creative contribution without requiring new theoretical frameworks.
Similarly, digital sampling cases offer additional precedent for analyzing technologically-mediated creative appropriation. In Bridgeport Music, Inc. v. Dimension Films (2005), Grand Upright Music, Ltd. v. Warner Bros. Records Inc. (1991), and more recently, VMG Salsoul, LLC v. Madonna Louise Ciccone (2016), courts have grappled with how existing doctrines apply to creative practices that use technology to extract, transform, and recontextualize protected materials.
More specific to fanart, cases like Lucasfilm Ltd. v. Shepperton Design Studios (UK) demonstrate the reach of copyright protection over fan-derivative outputs. In this case, a company producing replicas of Star Wars stormtrooper helmets was found to have infringed copyright—even though the helmets were reinterpreted objects rather than exact copies. The court held that the original design constituted a protectable artistic work, reinforcing the scope of IP control over fan-derived artifacts.
Platforms such as DeviantArt and Tumblr have periodically found themselves at the center of takedown disputes. Although these sites function as hubs for fan creativity, they are also subject to DMCA enforcement and automated copyright detection algorithms, which have occasionally resulted in the removal of artworks that arguably qualify as fair use or transformation. The development of user-generated content platforms has tested copyright's adaptability without breaking its conceptual foundations. Cases involving YouTube, fan fiction archives, and social media platforms have applied established principles of secondary liability, safe harbor provisions, and fair use to novel technological contexts. Such interventions underscore the precariousness of fanart's online visibility—tolerated until deemed threatening or commercial.
By contrast, some corporations have embraced a more permissive and even promotional approach to fan creativity. A notable example is Starbucks, which has actively encouraged customers to decorate its iconic paper cups and share their designs via social media using branded hashtags (Mussies, 2020a). The company has also periodically featured user-generated illustrations of its siren logo in digital campaigns and store displays, framing such acts of artistic engagement not as infringement, but as a form of brand co-creation. This strategy reflects an understanding of fanart as value-adding, rather than parasitic—a collaborative mode of cultural participation that enhances visibility and emotional affinity. However, this model depends heavily on corporate discretion and branding logic; the invitation to "play" with IP is extended only so long as it aligns with the company's marketing objectives. It is a gesture of conditional generosity, not a surrender of legal control.
These cases and historical precedents reveal both copyright law's capacity to evolve incrementally in response to technological innovation and a foundational dilemma: its focus on economic rights and original authorship is often ill-suited to hybrid, communal, or affect-driven modes of creation. While traditional fanart emerged within this ambiguous zone, it was, at least, tethered to human intentionality and labor. As subsequent sections will show, the introduction of generative AI has significantly altered this equation—both by changing the mode of production and by amplifying the legal and ethical stakes of derivative visual culture.
3. The Development of AI-Generated Fanart: Evolution and Disruption
3.1 Technological Context: From Diffusion Models to Visual Mimicry
The emergence of generative AI tools has significantly altered the mechanics of image production, including within fanart communities (Mussies, 2023). Rather than representing a paradigmatic break, these technologies can be understood as an evolution in the long history of computer-assisted creative tools—albeit one that introduces unprecedented questions of scale, accessibility, and authorial distance.
At the core of many of these systems are large language models (LLMs), such as GPT-4 or Claude, which are trained on extensive corpora of textual data to predict and generate human-like language. When paired with diffusion-based image generators, such as Stable Diffusion, Midjourney, or OpenAI's DALL·E, these systems operate through multimodal architectures, mapping linguistic inputs to visual outputs. Diffusion models are trained on massive datasets comprising text-image pairs and function by learning statistical correlations between verbal descriptions and pixel-level arrangements. Unlike earlier AI techniques focused on classification or segmentation, diffusion models invert the process: they begin with random noise and iteratively "denoise" it to match a desired output conditioned on the prompt.
These tools build upon decades of incremental progress in computational creativity. From the rule-based systems of the 1970s to the neural networks of the 1990s and the generative adversarial networks (GANs) of the 2010s, machine learning approaches to image creation have developed through continuous refinement. Earlier digital creation software already raised questions about the relationship between human creativity and technological assistance. Adobe Photoshop's Content-Aware Fill feature, introduced in 2010, used algorithmic processes to generate novel pixel arrangements based on surrounding image data, while 3D modeling software has long included procedural generation features.
To understand the mechanics behind this shift, it is helpful to examine how these systems actually generate images from textual prompts. The process involves far more than simple text interpretation—it comprises a cascade of neural computations, semantic mappings, and probabilistic sampling. The diagram below provides a technical overview of this pipeline.
Figure 1. Prompt-to-Image Pipeline: A logic-style block diagram illustrating how a Large Language Model (LLM), paired with a diffusion-based generator, transforms user prompts into images.
This visual framework reveals the multi-stage structure of systems like DALL·E or Stable Diffusion: from semantic parsing and embedding creation, through latent denoising, to final image refinement and delivery. Each layer of the diagram corresponds to a distinct computational function: from the initial linguistic processing of user input to the cross-modal alignment of textual and visual embeddings, and finally to the iterative diffusion stages that render synthetic imagery.
This technical process reveals important continuities with established digital creation methods. The "semantic parsing" stage parallels how human artists interpret conceptual briefs. The "cross-modal alignment" resembles how artists translate verbal concepts into visual compositions. The "iterative diffusion" process mimics the gradual refinement that occurs in traditional digital painting. What distinguishes AI generation is primarily the degree of automation, the statistical basis of creative choices, and the embedding of less visible components: training datasets scraped without consent, safety filters guided by proprietary heuristics, and generative configurations often shaped more by engineering constraint than by artistic sensitivity.
Through this iterative refinement, these systems can render novel images that stylistically and compositionally resemble the training data, not by reproducing specific examples, but by recombining aesthetic patterns derived from statistical learning. Yet, the line between inspiration and replication becomes increasingly porous, particularly when users engineer prompts to mimic specific franchises, visual aesthetics, or even individual artists. These capabilities have given rise to an explosion of AI-generated fanart, much of which closely resembles (and sometimes competes with) traditional fan-created imagery.
A recent and widely publicized case study in the ambiguous cultural and legal status of AI-generated fanart is the proliferation of "Ghibli-style" images produced via generative AI tools such as ChatGPT-4o. Fuelled by OpenAI's newly integrated visual interface, users rapidly began generating illustrations that mimicked the iconic visual style of Studio Ghibli—the hand-crafted, emotionally resonant aesthetic developed over decades by Hayao Miyazaki and his collaborators. These images draw from the same visual language that human digital artists have emulated for decades. However, rather than requiring years of study and practice, they can be produced in seconds with a well-crafted prompt.
From a legal standpoint, this phenomenon falls into an increasingly fraught category: stylistic pastiche. Under most copyright regimes, artistic style as such is not protected—it is ideas, not expressions, that are free. Yet when an AI-generated image effectively reproduces the "look and feel" of a known artist or franchise with high fidelity, a host of potential claims emerge: misappropriation of moral rights (in civil law jurisdictions), trade dress confusion (under US Lanham Act principles), and even unfair competition.
This debate is not abstract. As artist and Ghibli co-founder Hayao Miyazaki remarked in 2016, AI-generated animation is "an insult to life itself", a reaction not to technological progress, but to the devaluation of lived experience and intentional craft it can entail. His words have taken on renewed relevance as images of Ghibli-style caricatures - including, controversially, depictions of historical atrocities - circulate freely online, generated within seconds by users who may never have studied animation, visual storytelling, or Japanese culture. Critics such as Vaughan-Nichols (2025) argue this constitutes a form of "industrial copy-and-paste" theft: derivative in every sense but legal.
3.2 Illustrative Conflicts and Legal Applications
Recent jurisprudential developments illustrate the mounting tensions between artificial intelligence-generated artistic expressions and established intellectual property rights frameworks, particularly when such derivations approximate recognizable commercial intellectual properties. While these cases demonstrate novel applications, they can be analyzed using copyright's existing analytical frameworks for derivative works, substantial similarity, and transformative use.
In early 2024, SEGA Corporation initiated a series of Digital Millennium Copyright Act (DMCA) takedown notices targeting AI-generated imagery depicting the 'Sonic the Hedgehog' character in hyperrealistic and anime-inspired renditions. The legal significance of these enforcement actions lies in their application to works containing no direct reproduction of protected elements, but rather stylistic interpretations and character resemblances (Wadhwa & Smith, 2024). While the technological means of creation differs from manual drawing, the legal question remains consistent with established doctrine: whether the works contain substantial similarity to protected elements of SEGA's intellectual property. As Lord Justice Jacob noted in Nova Productions Ltd v Mazooma Games Ltd [2007] EWCA Civ 219, 'non-literal copying' may still constitute infringement when substantial parts of the original work are reproduced, even if transformed. This principle applies to both human-created and AI-generated fanart without requiring distinct analytical frameworks.
Similarly, Nintendo's response to Midjourney-generated portraits resembling characters from 'The Legend of Zelda' franchise offers another instructive example. Whilst Nintendo did not pursue formal litigation, it issued categorical statements affirming that unauthorized derivative works—including those facilitated through AI systems—constitute infringements of its intellectual property rights (Harrison & López, 2024). This episode exemplifies what legal scholars have termed the 'chilling effect' of corporate warnings on creative expression (Lessig, 2004). Under established principles articulated in cases like Campbell v Acuff-Rose Music, Inc 510 U.S. 569 (1994), the analysis turns not on the technological means of creation but on the relationship between the secondary work and the protected elements of the original. The distinction between transformative fair use and infringing derivative works remains particularly contentious in this context. The key questions—whether the work adds "something new, with a further purpose or different character" and whether it substitutes for the original in the marketplace—apply equivalently to human-created and AI-generated content, though the extent to which algorithmically-assisted fan creativity satisfies this standard remains subject to significant juridical debate (Samuelson, 2023).
AI-generated artwork reviving the pixelated aesthetic conventions of 1980s home computing platforms—such as the Commodore 64 or ZX Spectrum—presents particularly nuanced questions regarding intellectual property claims by dormant or dissolved software publishers. While foundational stylistic elements may reside in the public domain, specific sprite patterns, user interface components, and distinctive visual arrangements may retain copyright or trademark protection (Griffiths, 2023). As established in Norowzian v Arks Ltd [2000] EMLR 67, copyright protection extends to the expression of ideas rather than the ideas themselves, creating significant complexity when algorithmic systems reconstitute stylistic conventions without directly reproducing protected elements (Aplin & Davis, 2023). This distinction remains applicable regardless of whether the stylistic emulation occurs through human study or machine learning.
This phenomenon bears striking historical parallels to the software appropriation practices in the former German Democratic Republic, where developers created unauthorized replicas of Western computer games due to import restrictions and technological embargoes (Sterbenz, 2022). These historical precedents illustrate that contemporary tensions between intellectual property rights and derivative creation have significant historical antecedents, as documented in Mussies' (2020) autoethnographic account of amateur game development practices in the Netherlands during the mid-1990s. Mussies describes a proto-'produser community' engaged in transformative game modification using accessible development tools such as 'Klik & Play', demonstrating that participatory remix culture predates contemporary AI tools.
These exemplars demonstrate that whilst the fundamental principles of infringement remain largely consistent—unauthorized derivation, likelihood of confusion, and economic harm—the vectors through which potential violations occur have undergone significant transformation. Algorithmic systems now function as intermediaries in the creative process, and prompting has emerged as a de facto form of authorship, challenging traditional copyright frameworks that presuppose human creative agency (Ginsburg & Budiardjo, 2022). As Lord Hoffmann observed in Designer Guild Ltd v Russell Williams (Textiles) Ltd [2000] 1 WLR 2416, the determination of substantial similarity in copyright cases requires consideration of both the 'architecture of ideas' and their specific expression. This judicial framework faces unprecedented challenges when applied to AI-generated content, where the distinction between ideas and expression becomes increasingly indeterminate (Bently & Sherman, 2022).
3.3 Attribution, Authorship, and Algorithmic Agency: Evolving Frameworks
At the heart of the AI-fanart debate lies a fundamental legal and philosophical question: can an algorithm be an author? Most legal systems answer no—copyright protection is traditionally reserved for natural persons (or in some cases, legal persons such as corporations). Questions of attribution and authorship in AI-generated content can be effectively analyzed using copyright's existing frameworks for works involving technological mediation, though these frameworks are increasingly stretched by new technological capabilities.
In the UK, the Copyright, Designs and Patents Act 1988 explicitly states that for computer-generated works, "the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken." This provision offers a doctrinal foundation for analyzing prompt-based creation without requiring new theoretical constructs. The interpretive challenge lies not in developing new frameworks but in applying this established principle to determine which human contributions constitute "arrangements necessary" for creation. This provision suggests that prompt engineers or system designers may claim authorship—yet the boundaries of this attribution remain legally underdefined.
Prompt engineering's relationship to authorship parallels questions courts have previously addressed regarding conceptual art, appropriation practices, and directed photography. In cases like Andrien v. Southern Ocean County Chamber of Commerce (1983), courts have recognized that authorship can exist in the conception and direction of creative work even when the physical execution is performed by others. Similarly, in Burrow-Giles Lithographic Co. v. Sarony (1884), the creative selection and arrangement decisions underlying a photograph were deemed sufficient for authorship despite the mechanical nature of the camera's operation. These precedents provide analytical tools for evaluating prompt engineering's creative significance.
Moreover, the notion of originality—a cornerstone of copyright protection—becomes difficult to assess when the "creative choices" are made by statistical models trained on unlicensed, uncredited human artworks. Some scholars have proposed that prompting constitutes a form of creative curation, while others argue it is more akin to triggering a random number generator with aesthetic constraints. The attribution challenges posed by AI generation also have parallels in sampling, appropriation art, and digital collage practices. Courts have developed nuanced approaches to analyzing when technological extraction and recombination of existing works constitutes protectable new expression.
Regardless of doctrinal resolution, the practical effect is clear: AI-generated fanart occupies a space where traditional concepts of authorship, infringement, and moral rights are increasingly strained. This becomes especially problematic when AI systems replicate culturally specific artistic styles—such as manga or Ghibli aesthetics—without respecting their cultural, ethical, or historical embeddedness. What happens when the labor of fanart is decoupled from human craftsmanship? When aesthetic fluency can be simulated in seconds? And when culturally specific styles—once preserved by training, immersion, and situated knowledge—become prompts in a dataset?
The convergence of these technological developments and legal tensions invites a deeper examination of what is at stake—legally, ethically, and affectively—when style is extracted from substance, and when machines mimic, but do not understand, the traditions they replicate. As we explore in subsequent sections, these questions demand not only legal analysis but also ethical and philosophical consideration of the relationship between technology, creativity, and cultural heritage.
4. Doctrinal Analysis of Copyright Law: Continuity and Adaptation
As Giulia Bassini (2025) writes, the EU’s Regulation 2024/1689, also known as the AI Act, represents a major step toward harmonised and rights-based governance of artificial intelligence within the European Union. Entering into force in August 2024, the Act introduces a risk-based classification system that bans AI practices deemed to pose an unacceptable threat to individuals and society—such as biometric surveillance, predictive policing, and emotion recognition in education or workplaces—while tightly regulating so-called high-risk systems, including those used in healthcare, law enforcement, and critical infrastructure. Developers of these systems must adhere to strict obligations, such as conducting risk assessments, using unbiased datasets, ensuring transparency and traceability, and implementing human oversight and cybersecurity measures. Although widely praised for its ambition, the Act has sparked debate among scholars who raise concerns about conceptual vagueness, potential overregulation—particularly for small developers—and the challenge of enforcement across member states. Critics also question the Act’s extraterritorial scope and its ability to keep pace with the rapid evolution of AI technologies, suggesting that a more flexible, principle-based approach may ultimately prove more sustainable.
As AI-generated fanart continues to proliferate across digital platforms, legal systems worldwide are being pressed to confront a doctrinal challenge: how should copyright law respond when machines simulate human creativity—and do so by drawing upon, remixing, or restyling pre-existing cultural forms? This section examines the core legal questions raised by this phenomenon, with particular attention to jurisdictional differences between the United Kingdom, the European Union, the United States, and Japan.
At stake is not only the question of ownership, but the deeper structural issue of whether current legal categories—from originality to fair use—remain adequate in an era where artistic agency is dispersed across humans, machines, and data infrastructures. Through analysis of current doctrine, case law, and scholarly debate, I argue that copyright's existing frameworks demonstrate remarkable adaptability, requiring careful interpretation rather than wholesale reinvention.
4.1 Originality Standards Across Jurisdictions: Nuance and Application
The cornerstone of copyright protection in most legal systems is the principle of originality, though its interpretation varies significantly across jurisdictions. In the EU, following the Court of Justice's landmark decision in Infopaq International A/S v. Danske Dagblades Forening (CJEU 2009), originality requires a work to be the "author's own intellectual creation." However, this standard has been inconsistently applied across EU member states, with some jurisdictions like Germany traditionally requiring a higher threshold of "personal intellectual creation" (persönliche geistige Schöpfung), while others like France have applied the standard more liberally through their concept of l'originalité.
In the United Kingdom, while formally adopting the Infopaq standard following Brexit, courts have continued to apply it through the lens of the traditional UK approach requiring "skill, labor, and judgment" as seen in cases like Newspaper Licensing Agency v. Meltwater (2011). This has resulted in a hybrid approach that acknowledges the CJEU standard while preserving elements of the UK's historically lower threshold.
In the United States, the threshold established in Feist Publications v. Rural Telephone Service (499 U.S. 340 [1991]) requires that a work be "independently created" and possess "at least some minimal degree of creativity." This explicitly rejected the "sweat of the brow" doctrine and established a constitutionally mandated, albeit minimal, creativity requirement.
Japanese copyright law, while not adopting the exact terminology of these other jurisdictions, requires a work to be "creatively produced" (sōsakubutsu) and to express the thoughts or feelings of the author. This standard, as interpreted by Japanese courts, tends to focus on the presence of creative choices that reflect the author's personality.
These varying standards raise specific challenges for AI-generated works. The UK has attempted to address this through Section 9.3 of the Copyright, Designs and Patents Act 1988, which states that for computer-generated works where "there is no human author," the author is defined as "the person by whom the arrangements necessary for the creation of the work are undertaken." Contrary to some misinterpretations, UK courts have developed a consistent approach to identifying these human "arrangers" as authors—typically the individuals who select parameters, provide inputs, or curate outputs. In cases like Nova Productions v. Mazooma Games (2007), the court identified the programmer as the author of computer-generated images, establishing a precedent that has been followed in subsequent decisions.
Japan has taken a more explicit legislative approach. Recent amendments to Japanese copyright law in 2023 have specifically addressed AI outputs, establishing that while fully autonomous machine-generated content remains unprotected, human creators who substantively direct or modify AI outputs may claim authorship rights. These amendments reflect a careful balancing of traditional authorship concepts with technological realities.
The application of these originality standards to AI-generated fanart requires nuanced analysis rather than doctrinal reinvention. The key question across jurisdictions becomes: which aspects of human-machine collaboration satisfy existing originality thresholds? This question is not unprecedented but represents the latest evolution in copyright's ongoing adaptation to technological change.
4.2 Derivative Works Doctrine: Consistency Across Creative Methods
A second axis of legal consideration involves the status of derivative works. In traditional copyright doctrine, a derivative work is one based upon pre-existing works that recasts, transforms, or adapts them—such as translations, musical arrangements, or film adaptations.
The legal question becomes: does AI-generated fanart constitute a derivative work if it imitates, but does not directly copy, the original material? For instance, when a prompt engineer directs an AI system to generate an image "in the style of Studio Ghibli," does the resulting output infringe Ghibli's intellectual property rights?
In U.S. law, the determination hinges on the "substantial similarity" test—whether the AI output incorporates protected elements of the original work, regardless of the means of creation. Cases like Warner Bros. Entertainment Inc. v. RDR Books (2008) have established that even works created through significant transformation may constitute derivatives if they appropriate protected expression from the original. This doctrine applies equivalently to hand-drawn and AI-generated fanart without requiring distinct analytical frameworks.
In EU law, particularly under the InfoSoc Directive, the analysis focuses on whether the new work evokes the protected character of the original and whether applicable exceptions apply. The Court of Justice has consistently emphasized that the reproduction of even small portions of protected works may constitute infringement if those portions contain elements that express the author's intellectual creation, as established in cases like Pelham v. Hütter (2019).
Japanese law offers additional protection through its robust recognition of moral rights, which protect not only against unauthorized reproduction but also against actions that might damage the author's reputation or the integrity of their work. The Ryoichi Ikegami v. Shogakukan case demonstrated that stylistic imitation affecting the integrity of the original author's work may constitute infringement—a principle applicable across creative technologies.
The application of derivative work doctrine to prompt-engineered content represents not a doctrinal rupture but a familiar analytical challenge: determining whether specific prompts are designed to generate outputs that incorporate substantial protected expression from existing works. This analysis parallels questions courts have addressed regarding other technologies that facilitate creative adaptation.
4.3 Fair Use, Exceptions, and the Limits of Transformation
The applicability of fair use (US) and statutory exceptions (UK, EU, Japan) represents a third legal domain impacted by AI-generated content. These doctrines provide limited exemptions from copyright infringement while maintaining copyright's fundamental protections.
In U.S. law, the fair use doctrine is evaluated across four factors: (1) the purpose and character of the use, (2) the nature of the copyrighted work, (3) the amount and substantiality of the portion used, and (4) the effect on the potential market. Central to contemporary fair use analysis is the concept of "transformative use"—whether the new work "adds something new, with a further purpose or different character" as established in Campbell v. Acuff-Rose Music (510 U.S. 569 [1994]).
The Supreme Court's recent decision in Andy Warhol Foundation for the Visual Arts v. Goldsmith (2023) has refined this doctrine by emphasizing market substitution and commercial purpose. The Court held that Warhol's stylistic transformation of Goldsmith's photograph did not constitute fair use when used for the same commercial purpose as the original, despite aesthetic changes. This decision suggests that AI outputs that mimic pre-existing imagery—particularly for distribution or monetization—may face heightened scrutiny under fair use analysis.
At the same time, decisions like Cariou v. Prince (2013) demonstrate that significant transformation through technological means can constitute fair use when it creates new aesthetic meaning. This precedent provides a framework for analyzing AI-generated transformations that substantively recontextualize their inputs.
In the UK and EU, exceptions for caricature, parody, and pastiche provide more limited protection than U.S. fair use. Recent UK jurisprudence such as PPL v. ITV (2019) demonstrates a judicial preference for strict textual interpretation of these exceptions, limiting their application to clearly defined categories. Similarly, the CJEU has interpreted exceptions narrowly, as seen in cases like Pelham v. Hütter (2019), where the Court rejected applying the quotation exception to music sampling.
Japanese copyright law provides specific exceptions for private use, educational purposes, and certain forms of quotation, but these are narrowly construed and may offer limited protection for AI-generated fanart. Japan's emphasis on moral rights further restricts the scope of permissible transformation without authorization.
The application of these exceptions to AI-generated content turns not on developing new doctrinal frameworks but on carefully analyzing whether prompt engineering constitutes sufficient creative transformation to qualify for existing exceptions. This inquiry parallels questions courts have addressed regarding other technologically facilitated creative practices.
4.4 Critical Analysis of Judicial and Scholarly Approaches
Recent jurisprudence and scholarly discourse demonstrate copyright's engagement with technological change through incremental doctrinal refinement rather than paradigmatic shifts. These developments provide valuable guidance for analyzing AI-generated content.
The Warhol v. Goldsmith decision illustrates copyright's capacity to address new forms of artistic appropriation through established doctrinal tools. The Court's emphasis on purpose and market rather than merely aesthetic transformation offers a framework applicable to various technologies of reproduction and adaptation.
European jurisprudence similarly demonstrates adaptation to technological change through principled application of existing doctrine. The CJEU's decision in cases like Pelham v. Hütter addressed digital sampling by applying established principles of reproduction and exception, providing a framework for analyzing various forms of technological extraction and transformation.
Japanese scholarly discourse on AI authorship has evolved significantly, particularly following the 2023 copyright amendments. Scholars like Nakagawa and Saito have moved beyond debating whether AI outputs can receive protection to exploring the more nuanced question of how human-machine collaboration affects traditional authorship concepts. This discourse reflects not a rejection of existing frameworks but their thoughtful application to new technological contexts.
From a theoretical perspective, legal scholars have increasingly examined the implicit assumptions underlying copyright's approach to machine-mediated creativity. Rebecca Tushnet (2022) has critiqued inconsistent application of the transformative test when human agency is diffused across technological systems. Amanda Levendowski (2020) has advocated recognizing curatorial labor as a form of authorship—a position that could extend protection to sophisticated prompt engineering. Anupam Chander (2023) has warned against extending copyright to AI-generated outputs without careful consideration of market effects and public domain interests.
These scholarly perspectives reveal that the fundamental challenge is not developing entirely new copyright doctrines but refining existing frameworks to accommodate the complex realities of human-machine collaboration. Copyright law has historically evolved through incremental adaptation to technological change—from photography to software to digital sampling—and AI-generated content represents the continuation of this evolutionary process rather than a revolutionary break.
The task for courts and policymakers is not to discard copyright's foundational principles but to interpret them with sensitivity to both technological reality and copyright's underlying purposes. This requires nuanced attention to the specific human contributions involved in AI-generated content—from prompt engineering to system selection to output curation—and their relationship to copyright's traditional concepts of originality, transformation, and authorship.
5. Human-Machine Collaboration: Legal and Ethical Implications in Historical Context
The emergence of AI-generated fanart necessitates a careful examination of human-machine collaboration within creative processes. Rather than representing an unprecedented disruption, AI tools function as the latest in a series of technological mediators that transform human creative intent. This section analyzes both the novelty and continuity in these collaborative processes, addressing their legal and ethical implications through historically grounded frameworks.
5.1 Prompt Engineering: Contextualizing a New Form of Creative Direction
A central question in both copyright doctrine and fan studies is how to characterize prompt engineering as a creative act. As documented in earlier work (Mussies, 2023), fans engage with text-based prompts to produce visual renderings of familiar characters that occupy an interpretive space between original creation and derivative work.
This practice can be productively understood through historical analogies to other forms of indirect creative control. Conceptual artists like Sol LeWitt created works by writing instructions for others to execute, emphasizing conception over manual execution. His "Wall Drawings" consisted of written directives that others implemented according to specified parameters—a process analogous to prompt engineering's use of textual instructions to generate visual outputs. Similarly, architectural design has long involved architects creating specifications that others transform into physical structures, with copyright law recognizing such works as protectable expression despite their collaborative realization.
These precedents suggest that prompt engineering can represent a meaningful form of creative expression when it involves sufficient specificity, intentionality, and aesthetic judgment. However, this recognition must be balanced against important distinctions. Unlike conceptual artists or architects, prompt engineers operate within parameters established by model developers and have limited control over how their instructions are interpreted. The resulting outputs emerge from probabilistic processes rather than deterministic execution.
This creates an important nuance in how we understand prompt engineering's relationship to authorship. When prompts are generic ("create a fantasy landscape") or primarily reference existing properties ("draw in Ghibli style"), they may not rise to the threshold of original expression. However, when prompts demonstrate substantial creativity in their construction—combining elements in innovative ways, specifying unique compositional details, or establishing novel aesthetic parameters—they may constitute a form of authorship within existing doctrinal frameworks.
This position resolves the apparent contradiction between recognizing prompt engineering as potentially meaningful authorship while acknowledging the stochastic nature of AI generation. The resolution lies in recognizing that prompt engineering exists on a spectrum: from minimal creative input to substantial creative direction, with legal recognition properly reserved for instances where human creative judgment substantially shapes the resulting work. This approach allows for a nuanced assessment that neither dismisses all prompt engineering as mechanistic activation nor elevates all prompting to the status of original authorship.
5.2 Integration Practices: Beyond the Binary of Input and Output
The relationship between AI outputs and creative practice is more complex than simple generation. As observed in fan communities, AI-generated images rarely function as final artifacts but instead serve as components in broader creative processes. An AI-generated image might inspire a hand-drawn redesign; a Midjourney portrait might serve as visual reference material for character development in fanfiction; a latent texture might be isolated, recolored, and reassembled into an entirely new composition.
This integration challenges the binary distinction between "input" and "output" that dominates technical and legal discourse around AI generation. When fans incorporate AI outputs into multistage creative processes, they transform what might initially be derivative or unoriginal elements into components of more complex creative works. This practice has precedents in collage, sampling, and other forms of creative appropriation that copyright law has developed frameworks to address.
The integration of AI outputs into broader creative practices also raises important questions about platform governance and ownership. Who owns the AI-generated image: the user, the platform, the model developer? And what happens when a fan's prompt causes the model to reproduce recognizable elements of another's fanart, or even an original work buried deep in the training set?
These questions can be addressed through established frameworks for analyzing intermediary liability and platform responsibility. The challenge of balancing creator rights with platform interests has precedents in earlier digital contexts, from photography sharing sites to music distribution platforms. While AI generation introduces new technical complexities, the fundamental questions of control, consent, and attribution remain conceptually similar to those addressed in other technological contexts.
What distinguishes AI platforms is primarily the degree of abstraction between original works and derivative outputs, rather than a fundamental shift in legal relationships. This abstraction creates practical challenges for attribution and consent but can be analyzed through established doctrinal tools rather than requiring entirely new frameworks.
5.3 Cultural Appropriation and Style: Toward a Balanced Framework
The question of style appropriation through AI presents one of the most challenging aspects of human-machine collaboration. When AI systems produce "Ghibli-style" or "manga-inspired" imagery based on training data that includes works from specific cultural traditions, they raise important questions about cultural respect, attribution, and context.
These concerns parallel broader debates about cultural appropriation in creative practice. However, a balanced approach must recognize two important principles that initially appear contradictory but can be reconciled:
First, style itself contains important cultural and artistic significance that merits respect and attribution. When AI systems reproduce stylistic elements associated with specific cultural traditions or individual artists without acknowledgment, they risk decontextualizing aesthetic forms from their historical, cultural, and ethical grounding. This concern applies equally to human fanartists who appropriate cultural elements without contextual understanding, suggesting that the issue lies not with the technology itself but with how cultural borrowing occurs across contexts.
Second, copyright law has historically avoided extending protection to style alone, recognizing that overly broad style protection would create insurmountable barriers to creative expression. This limitation serves important public interest goals by ensuring that aesthetic vocabularies remain available for creative reuse and reinterpretation.
This apparent tension can be resolved through a framework that distinguishes between legal protection and ethical practice. While copyright may properly refrain from extending formal protection to style alone, ethical creative practice—whether human or machine-assisted—should incorporate appropriate attribution, contextual awareness, and respect for cultural origins. This approach allows for creative exchange while avoiding exploitative appropriation.
The solution lies not in expanding copyright to protect style—which would indeed create excessive barriers to expression—but in developing norms, practices, and potentially non-copyright legal mechanisms (such as attribution requirements) that acknowledge stylistic lineage without prohibiting creative engagement. Such an approach respects the cultural significance of style while preserving the creative freedom that copyright's limitations on style protection are designed to enable.
5.4 Labor Value and Creative Economics
The economic implications of AI-generated fanart demand careful consideration. If AI can generate art in milliseconds, what value remains for human artists—particularly those who rely on commissions, patronage, or platform visibility? This concern reflects broader questions about how technological change affects creative labor markets.
Historical perspective offers important context. Previous technological developments—from photography to digital design tools—have transformed creative labor without eliminating its value. Each innovation has displaced certain forms of artistic labor while creating new opportunities and forms of value. The societal valuation of human creativity has persisted despite technological change because it reflects deeper values than mere technical reproduction.
This historical pattern suggests that while AI generation will transform certain aspects of creative labor, it is unlikely to eliminate the distinctive value of human artistic expression. Several factors support this view:
First, as documented in fan communities, AI outputs often serve as components in broader creative processes rather than final products. This integration suggests that AI may function more as a tool that augments human creativity than as a replacement for it.
Second, the value of creative works derives not only from their visual qualities but from their connection to human experience, intention, and meaning-making. These qualities are not readily replicable through algorithmic processes alone.
Third, creative communities often value provenance, process, and relationship as much as final outcomes. The knowledge that a work emerged from human experience and intention creates forms of value that machine generation alone cannot replicate.
This perspective offers a more nuanced understanding than either technological utopianism or deterministic pessimism. It acknowledges that AI will significantly impact creative labor markets while recognizing that human creativity maintains distinct forms of value that technological reproduction alone cannot displace.
5.5 Toward Integrated Understanding
The relationship between human creativity and machine generation is neither one of simple replacement nor mere augmentation. It represents a complex reconfiguration of creative processes that both extends and challenges existing frameworks for understanding authorship, ownership, and cultural exchange.
Legal analysis of these relationships benefits from both historical perspective and direct engagement with emerging practices. While AI-generated fanart raises important questions about authorship, derivation, and cultural respect, these questions can be addressed through thoughtful application and incremental adaptation of existing doctrinal frameworks rather than wholesale reinvention.
The key insight is that human-machine collaboration in creative contexts represents evolution rather than revolution—the latest development in a long history of technologically mediated creativity that has consistently challenged and refined our understanding of authorship, originality, and creative value. By recognizing both the continuities and the novel aspects of AI-assisted creation, we can develop legal and ethical frameworks that protect important creative interests while enabling innovation and cultural exchange.
6. Policy Considerations: Balanced Evolution for Human-Machine Creativity
As AI-generated fanart increasingly occupies the interstices of legality, creativity, and automation, there is growing recognition that intellectual property regimes require thoughtful adaptation to address the complexities of human-machine collaboration. This section examines current regulatory developments, proposes balanced policy recommendations, and argues for an approach that combines doctrinal refinement with targeted interventions—preserving copyright's core functions while addressing emerging challenges.
6.1 Doctrinal Refinement: Evolution Within Existing Frameworks
Current copyright frameworks are premised on concepts of human originality, individual authorship, and expressive autonomy—all of which are challenged by generative AI systems. However, rather than requiring wholesale reinvention, these challenges can be addressed through thoughtful doctrinal refinement that builds upon copyright's historical capacity for adaptation.
The most promising approach involves interpreting and extending existing legal categories to accommodate new forms of creative collaboration. For instance, courts could develop clearer interpretive guidance for applying the "arrangements necessary" standard in the UK's computer-generated works provision (CDPA 1988, Section 9.3) to determine which aspects of prompt engineering constitute meaningful authorial contribution. This would provide greater certainty without requiring entirely new legislative categories.
Similarly, originality analysis can evolve to recognize creative choices that operate at different levels of abstraction—from system selection and prompt crafting to output curation and modification. This evolution would extend established originality principles rather than replacing them with AI-specific standards. In jurisdictions like the United States, the flexibility of the "minimal creative spark" standard established in Feist provides doctrinal space for recognizing prompt engineering as potentially satisfying originality requirements when it involves substantial creative judgment.
Derivative work doctrine can likewise be refined to clarify when prompt engineering constitutes an attempt to reproduce protected expression rather than merely reference unprotectable style or genre conventions. This clarification would build upon existing distinctions between protected expression and unprotected ideas—a fundamental copyright principle that remains relevant in the AI context.
These refinements represent evolution within copyright's established conceptual framework rather than paradigmatic shifts. They acknowledge technological change while maintaining doctrinal continuity and predictability—values essential to a functioning copyright system.
6.2 Regulatory Developments and Approaches
Recent regulatory consultations suggest that legal systems are beginning to engage more seriously with the challenges posed by AI-generated content. In the UK Intellectual Property Office’s 2022 consultation on AI and IP, respondents expressed divergent views. Some advocated for a “creator-neutral” model, extending authorship to AI outputs, while others warned that such changes could undermine human creators and disrupt artistic labour markets. The UK government ultimately decided that AI-generated works would not receive copyright protection in the absence of human authorship, although it remains open to revisiting this stance as technologies and creative practices continue to evolve.
At the international level, the World Intellectual Property Organization (WIPO) has convened a series of stakeholder dialogues exploring the intersection of AI and IP. These sessions have surfaced tensions between technological innovation, cultural equity, and the enforcement of rights. Particularly striking were the concerns voiced by delegations from the Global South, who noted that AI systems trained on unlicensed content may disproportionately extract from marginalised or underrepresented creative communities. Their interventions raise important questions about global fairness and representation in AI development.
Japan, whose Copyright Act places strong emphasis on moral rights and the protection of individual expression, offers yet another regulatory approach. The Cultural Affairs Agency there has issued interpretive guidance that addresses the complexities of AI-generated content. This guidance seeks to balance recognition of AI's creative capabilities with a firm commitment to protecting human creators—especially in cases involving unauthorised stylistic imitation.
Across these jurisdictions, a growing consensus appears to be emerging: targeted regulatory interventions, rather than sweeping overhauls, offer the most promising path forward. Among the most frequently proposed measures are transparency requirements for AI training data, including documentation of data sources and opt-out mechanisms for creators; platform governance standards for AI-generated content, such as clear attribution practices and disclosure obligations; and educational initiatives to improve copyright literacy among AI users, helping to reduce inadvertent infringement. While these strategies acknowledge the unique complexities of AI, they also build on long-standing policy responses to technological change in the creative industries.
6.3 Stakeholder Recommendations: A Collaborative Approach
The path towards a more equitable and responsive copyright framework for AI-generated content demands the coordinated engagement of multiple stakeholders. Each plays a distinct role in shaping a legal and ethical ecosystem that fosters innovation while safeguarding legitimate creative interests.
For lawmakers, the imperative lies not in sweeping legislative reform, but in measured, incremental refinements that reflect both technological realities and doctrinal coherence. Statutory provisions regarding computer-generated works require clarification—particularly concerning the threshold of human input necessary to constitute authorship. Likewise, doctrines pertaining to derivative works and fair use (or fair dealing) must be recalibrated to address the novel complexities introduced by generative AI. In jurisdictions where moral rights enjoy strong protection, it is increasingly urgent to consider how such rights apply to works that emulate the distinctive style of particular artists, or that appropriate culturally specific traditions. The legal challenge here is to navigate between overprotection—which may stifle legitimate creative borrowing—and underprotection, which risks eroding the dignity and identity of original creators.
Digital platforms, as both facilitators and gatekeepers of AI-generated content, occupy a pivotal position in the governance landscape. Their role should extend beyond mere content hosting, towards the proactive implementation of transparency mechanisms and ethical oversight. Platforms ought to require explicit disclosure of AI involvement upon content upload, supported by robust metadata systems that inform users whether and how generative tools were employed. In tandem, clear terms of service must articulate the ownership status, reuse permissions, and attribution obligations attached to such works. Dispute resolution mechanisms should also be instituted to address concerns relating to stylistic imitation or misattribution, drawing upon and extending existing governance frameworks.
Finally, creators and fan communities must not be overlooked in this conversation, for they form the cultural lifeblood of the spaces where AI-generated art circulates and acquires meaning. These communities are well-positioned to articulate informal ethical standards, particularly with respect to the attribution of AI assistance and the responsible emulation of artistic styles. They can develop grassroots norms for respectful engagement with culturally embedded aesthetics, and contribute meaningfully to policy formation by participating in platform dialogues and advisory processes. In doing so, they offer a bottom-up complement to top-down regulation—grounding abstract legal principles in lived creative practice.
Only through such a pluralistic and dialogical approach can the legal system evolve in a way that honours the complexity of contemporary cultural production, while remaining faithful to the foundational values of authorship, expression, and justice.
6.4 Implementation Framework: A Balanced Matrix
Addressing the challenges posed by AI-generated fanart requires coordinated action across multiple domains. The following matrix outlines a balanced implementation framework that combines doctrinal refinement with targeted interventions:
This framework recognizes that effective governance requires coordination across legal, technical, and cultural domains. Rather than positioning AI-generated content as requiring entirely new regulatory paradigms, it emphasizes balanced adaptation of existing frameworks to address specific challenges while preserving copyright's core functions.
6.5 Conclusion: Toward Principled Evolution
The emergence of AI-generated fanart presents significant challenges to traditional copyright frameworks, but these challenges can be addressed through principled evolution rather than radical reinvention. By building upon copyright's historical capacity for adaptation to technological change, policymakers can develop approaches that accommodate human-machine collaboration while preserving essential creative and cultural values.The most promising path forward combines doctrinal refinement with targeted interventions—clarifying how existing legal principles apply to new creative practices while addressing specific concerns through tailored regulatory measures. This approach maintains legal predictability while acknowledging the unique characteristics of AI-generated content.
Ultimately, the goal should be a copyright system that continues to fulfill its fundamental purpose: encouraging creative expression while ensuring that creators receive appropriate recognition and compensation for their work. By approaching AI-generated content with both flexibility and principle, we can adapt copyright for the digital age without sacrificing its essential values.
7. Reflections: Artistic Authenticity, Style, and Legal Evolution
The conceptual challenge posed by AI-generated fanart necessitates a critical examination of both technological novelty and legal continuity. These artefacts exist at the intersection of established copyright principles and emerging technological paradigms, requiring a nuanced approach that neither overstates their revolutionary nature nor diminishes their distinct characteristics.
7.1 Historical Continuity and Technological Evolution
The ontological questions raised by AI generation have significant precedents in earlier technological and artistic developments. Photography's emergence in the nineteenth century prompted similar deliberations regarding mechanical mediation and creative agency, yet jurisprudence ultimately recognised that human creative choices in composition, arrangement, and selection could constitute authorship despite the camera's mechanical operation. Similarly, digital sampling raised questions about fragmentation and recombination of creative works that presaged current debates about training data and output generation. Conceptual art challenged traditional understandings of artistic execution by separating conception from physical creation—a precedent particularly relevant to analysing the prompt engineer's creative role.
These historical parallels suggest that whilst AI generation introduces new complexities, the fundamental questions it raises about authorship, originality, and creative transformation have precedents within copyright's evolution. The legal system's demonstrated capacity to adapt to earlier technological challenges suggests that existing doctrinal tools—thoughtfully refined and contextually applied—can accommodate AI-generated content without requiring paradigmatic shifts in legal theory.
7.2 AI Fanart as Cultural Diagnostic
Beyond legal categorisation, AI-generated fanart functions as what might be termed a cultural diagnostic—revealing the deep structures of aesthetic saturation, genre conventions, and symbolic recombination that define our post-digital media ecology. Drawing on Baudrillard's theory of the simulacrum, one might argue that these outputs do not merely imitate reality but simulate simulations—artefacts that illuminate the layers of cultural memory and stylistic citation that populate contemporary media ecosystems. This perspective treats generative AI not simply as a creative tool but as a revelatory medium that renders visible the accumulated patterns of cultural production.
This analytical frame helps reconcile the apparent contradiction between valuing human fanart whilst critiquing AI generation. Whereas human fanart, even when appropriating cultural elements without full contextual understanding, represents an intentional act of cultural dialogue—an embodied engagement with aesthetic traditions—AI systems aestheticise without understanding. The distinction lies not in the surface appearance of the works but in their ontological status: human fanart, however derivative, emerges from authentic cultural participation, whilst AI outputs represent statistical aggregations of formal patterns divorced from lived experience or intentional meaning-making.
7.3 The Problematic Status of Prompt Engineering
The increasingly popular claim that prompt engineers function as co-authors of AI outputs warrants critical scrutiny. Whilst this position appeals to intuitions of effort, judgement, and intent, it risks conflating activation with authorship. The generative model, however sophisticated, possesses neither intentionality nor aesthetic consciousness—it operates through probabilistic correlation rather than creative engagement. This does not diminish the significance of human involvement but locates authorial contribution primarily in subsequent acts of selection, modification, and contextual framing rather than initial prompting.
However, this position requires qualification to avoid internal contradiction. If we acknowledge that human creativity encompasses selection and curation—not merely ex nihilo creation—then prompt engineering does represent a form of creative expression, albeit one fundamentally different from traditional authorship. The creative contribution lies not in generating the output directly but in navigating the latent space of possibilities through strategic prompting and iterative refinement. This represents a form of mediated creativity that challenges traditional authorial paradigms without rendering the concept of authorship entirely obsolete.
7.4 Style Protection and Cultural Expression
The tension between protecting distinctive styles and enabling cultural expression presents one of the most profound challenges in this domain. The distinctive visual language of major creative franchises—whether Ghibli's atmospheric landscapes or Nintendo's character designs—constitutes a significant cultural and commercial asset deserving of protection. Yet comprehensive style protection would indeed create insurmountable barriers to expression, potentially monopolising aesthetic approaches that should remain in the cultural commons.
This apparent contradiction can be resolved through a more nuanced approach to style protection that distinguishes between distinctive stylistic signatures and broader aesthetic traditions. Legal protection should focus on specific combinations of stylistic elements that constitute recognisable creative signatures rather than broader aesthetic approaches or techniques. This calibrated approach would protect creators' distinctive expressions whilst preserving the cultural commons necessary for artistic dialogue and evolution.
7.5 Economic Substitution and Market Harm
The theoretical tensions surrounding AI-generated fanart become particularly acute when such works circulate within commercial ecosystems. When a fan employs generative technologies to produce imagery in the unmistakable style of a well-known franchise, and this image is subsequently commercialised, the output may function as an economic substitute for officially licensed work regardless of its technical legal status. Here, the discourse must shift from questions of authorship and originality to those of market harm and substitution.
This economic dimension highlights the limits of purely aesthetic or philosophical analysis. Where AI-generated content is indistinguishable from—and financially competitive with—traditional human production, the stakes become not only artistic or doctrinal but economic. This presents courts with a challenge that extends beyond copyright's traditional focus on formal similarity to questions of functional substitution and market impact—issues that require empirical analysis alongside doctrinal interpretation.
7.6 Towards an Integrated Framework
The complex challenges posed by AI-generated fanart necessitate an integrated analytical framework that accommodates both technological innovation and legal continuity. Rather than representing a paradigmatic shift requiring entirely new theoretical constructs, these works occupy a position within copyright's evolutionary trajectory—challenging established categories whilst remaining analysable through refined application of existing principles.
This perspective does not diminish the importance of careful analysis and targeted adaptation. AI's scale, accessibility, and opacity create practical challenges for copyright enforcement and attribution that require thoughtful response. The statistical nature of machine learning raises questions about the relationship between training data and output that demand nuanced application of established doctrines. Yet these challenges represent evolution rather than revolution—an intensification and recombination of familiar questions rather than unprecedented legal problems requiring entirely new frameworks.
This integrated approach resolves the apparent contradictions identified in earlier analyses by recognising that AI-generated fanart simultaneously represents both continuity and disruption. The appropriate response lies not in wholesale rejection or unconditional embrace of these technologies, but in careful calibration of existing legal principles to address their distinct characteristics whilst preserving the fundamental values of copyright: encouraging creative expression, protecting legitimate commercial interests, and enriching the cultural commons.
Conclusion
The advent of AI-generated fanart compels us to reimagine the foundations of authorship, originality, and legal protection in ways that strain the doctrinal seams of existing copyright law. As this paper has argued, the challenge is not merely technical, but conceptual: a recalibration of the authorial paradigm is necessitated by the emergence of hybridised creative acts, wherein the human and machine co-produce cultural artefacts through layered processes of prompt, algorithm, and aesthetic iteration.
Rather than positing AI as a radical rupture, this inquiry has situated it along a historical continuum of technological interventions in artistic production. From early photography to digital collage, the law has repeatedly confronted the question of how far it can stretch to accommodate novel means of expression without forfeiting its normative commitments. AI, however, tests this elasticity more forcefully than its predecessors, not least because its outputs often mimic—at scale and with uncanny fidelity—the style, substance, and identity of copyrighted source material.
What emerges is a field of contestation where the legal system must simultaneously safeguard the rights of original creators, uphold the integrity of authorship doctrines, and foster innovation in a participatory, digitally mediated culture. The tensions surrounding AI-generated fanart—particularly in relation to derivative work, fair use, and moral rights—are emblematic of broader shifts in the cultural economy, wherein creativity is increasingly distributed, collaborative, and computationally embedded.
Thus, the path forward lies not in the wholesale rejection or uncritical embrace of AI authorship, but in a nuanced re-articulation of legal categories. This paper has recommended targeted doctrinal refinements—grounded in comparative legal analysis and informed by philosophical considerations of agency and intent—that recognise the hybrid nature of human-machine creativity without collapsing the distinction between the tool and the creator. If copyright is to remain both relevant and just, it must evolve in ways that acknowledge the layered ontologies of authorship now unfolding across the digital continuum.
In this sense, AI-generated fanart does not simply pose a legal problem; it reveals the metaphysical fault lines of authorship itself. The challenge, then, is to craft a jurisprudence capacious enough to reflect the fluidity of contemporary cultural production—while remaining anchored in principles that preserve the human dignity and expressive freedom at the heart of intellectual property.
References
Aplin, T., & Davis, J. (2023). Intellectual Property Law: Text, Cases, and Materials (5th ed.). Oxford University Press.
Baker v. Selden, 101 U.S. 99 (1879). https://supreme.justia.com/cases/federal/us/101/99/
Bassini, G. (2025). Lights and Shadows of The First-Ever Legal Framework on AI Worldwide. The International Journal for Crime, Law, and AI. https://www.intjournalcrimelawai.com/lights-and-shadows-of-the-first-ever-legal-framework-on-ai-worldwide
Bently, L., & Sherman, B. (2022). Intellectual Property Law (6th ed.). Oxford University Press.
Bridgeport Music, Inc. v. Dimension Films, 410 F.3d 792 (6th Cir. 2005).
Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53 (1884).
Capgemini (2019). Capgemini announces Project FARM. https://www.capgemini.com/be-en/news/press-releases/capgemini-announces-project-farm/
Copyright Act of 1976, 17 U.S.C. §§ 101-810 (1976).
Grand Upright Music, Ltd. v. Warner Bros. Records Inc., 780 F.Supp. 182 (S.D.N.Y. 1991).
Ginsburg, J. C., & Budiardjo, L. A. (2022). Authors and Machines: The Legal Structures of Technological Art. Columbia Journal of Law & the Arts, 45(2), 131-198.
Griffiths, J. (2023). Preserving the Past: Copyright Protection for Abandoned Software. European Intellectual Property Review, 45(8), 497-512.
Cornish, W., Llewelyn, D., & Aplin, T. (2023). Intellectual Property: Patents, Copyright, Trade Marks and Allied Rights (10th ed.). Sweet & Maxwell.
Hautsch, J. (2022). Pic sets, fan cognition, and fannish networks of meaning. Transformative Works and Cultures, 38. https://doi.org/10.3983/twc.2022.2151
Harrison, F., & López, M. (2024). The Legend of Legal Legends: Nintendo's IP Enforcement Strategies. Entertainment Law Review, 35(3), 86-92.
Jenkins, H. (2006). Convergence culture: Where old and new media collide. New York University Press.
Lessig, L. (2004). Free Culture: How Big Media Uses Technology and the Law to Lock Down Culture and Control Creativity. Penguin Press.
Lucasfilm Ltd. v. Shepperton Design Studios [2011] UKSC 39.
Mussies, M. (2020a). The Siren's Lure - A Starbucks influencer marketing case study. Better Marketing.
https://www.academia.edu/42754849/The_Sirens_Lure_A_Starbucks_Influencer_Marketing_Case_Study
Mussies, M. (2020b). Recreating two of my 1994 homebrew games. Interactive Pasts. https://interactivepasts.com/recreating-two-of-my-1994-homebrew-games/
Mussies, M. (2023). Artificial Intelligence and the production of fan art. Transformative Works and Cultures, 40. https://doi.org/10.3983/twc.2023.2431
Mussies, M. (2025). Stalked by Code: Navigating the Psychological and Legal Landscapes of Online Harassment. The International Journal for Crime, Law, and AI. https://www.intjournalcrimelawai.com/stalked-by-code-navigating-the-psychological-and-legal-landscapes-of-online-harassment
Samuelson, P. (2023). AI Challenges to Copyright in the Courts. Berkeley Technology Law Journal, 38(1), 1-58.
Statute of Anne, 8 Ann. c. 21 (1710). [An act for the encouragement of learning by vesting the copies of printed books in the authors or purchasers of such copies, during the times therein mentioned].
Sterbenz, C. (2022). Digital Piracy Behind the Iron Curtain: Software Appropriation in the German Democratic Republic. Journal of the History of Computing, 44(2), 89-113.
Vaughan-Nichols, S. J. (2025, April 14). Miyazaki, AI and intellectual property. The Register. https://www.theregister.com/AMP/2025/04/14/miyazaki_ai_and_intellectual_property/
VMG Salsoul, LLC v. Madonna Louise Ciccone, 824 F.3d 871 (9th Cir. 2016).
Wadhwa, R., & Smith, A. (2024). DMCA Takedowns for AI-Generated Fan Art: A Legal Analysis. Journal of Intellectual Property Law & Practice, 19(4), 210-225.
By Martine Mussies
Martine Mussies is an artistic researcher and autistic academic based in Utrecht, the Netherlands. She is a PhD candidate at the Centre for Gender and Diversity at Maastricht University, where she is writing her dissertation on The Cyborg Mermaid. Martine is also part of SCANNER, a research consortium aimed at closing the knowledge gap on sex differences in autistic traits. In her #KingAlfred project, she explores the online afterlives of King Alfred the Great, and she is currently working to establish a Centre for Asia Studies in her hometown of Utrecht. Beyond academia, Martine is a musician, budoka, and visual artist. Her interdisciplinary interests include Asia Studies, autism, cyborgs, fan art and fanfiction, gaming, medievalisms, mermaids, music(ology), neuropsychology, karate, King Alfred, and science fiction. More at: www.martinemussies.nl and LinkedIn.
Disclaimer: The International Platform for Crime, Law, and AI is committed to fostering academic freedom and open discourse. The views and opinions expressed in published articles are solely those of the authors and do not necessarily reflect the views of the journal, its editorial team, or its affiliates. We encourage diverse perspectives and critical discussions while upholding academic integrity and respect for all viewpoints.