The end of artistic innocence - AI art authorship
- Astrea Nicodemo
- Mar 3
- 11 min read
Astrea Nicodemo
The Prompt or the Product? The Essential Question in AI Art Authorship
"Good artists copy; great artists steal."
The quote, mistakenly attributed to Picasso by Steve Jobs, actually has a long history of different interpretations and has survived because it reveals an uncomfortable backstory: "art has always been recursive". Each masterpiece carries with it the ghosts of previous forms. Innovation rarely comes out of nowhere; rather, it is a recombination, mutation, or recontextualization.
Behind the beauty of the works of art lies something more disturbing and complex. When the visionary is powerful, the appropriation of his art by other artists becomes "influence". When the visionary is unknown, his authorship can be totally erased. Thus, what begins as influence turns into embezzlement.
The difference is not the copy, but the power and authority.
Copying from art history
The Renaissance was based on imitation. It was the rebirth of something that had already existed. Michelangelo obsessively studied Greco-Roman sculpture, absorbing proportions, musculature and compositional harmony. His David was not an isolated miracle, but a transformation of classical ideals into Christian humanism. Caravaggio did not invent his dramatic light out of thin air. He intensified the light borrowed from his Venetian predecessors, translating it into works of extraordinary beauty, unique in their kind.
The influence was not shameful. It was structural.
In the twentieth century, appropriation became an explicit strategy. When Marcel Duchamp presented Fountain in 1917, he declared that the selection itself could constitute authorship. Production was no longer necessary. The artistic act had transformed from a manual to a conceptual one.
Pablo Picasso's Les Demoiselles d'Avignon (1907), often hailed as a radical break in Western art, indisputably incorporates the visual language of African masks. Yet African sculptors, whose formal innovations contributed to the birth of Cubism, remained anonymous in art history for decades. Picasso became “The Genius”. The cultures at the origin of his disruptive works became "inspiration and influence".
In 1981, Sherrie Levine re-photographed Walker Evans' images from the Great Depression era and exhibited them as her own works (After Walker Evans). The art world did not dismiss her as a plagiarist. On the contrary, the work became a milestone of postmodern criticism, questioning originality, authorship, and the myth of the solitary genius. Levine demonstrated that context can transform the copy into conceptual art.
The history of art reveals a model. Influence is celebrated when it comes from an authoritative source. It becomes embezzlement when the power is held by those who draw influence from a weak source.
The fashion industry and systemic copying
Fashion makes the mechanism even more explicit.
Often, the cycles of fashion are compared to a garbage bag that is turned upside down every x years, and the garment that appears first in the pile becomes the trend. This is not really the case, but the influence of specific historical periods or cultural movements is fundamental to determining the style. Fashion is always inspired by something. Very often it is inspired by itself and by the previous eras that have made its fortune or ruin.
In the 80s and 90s, designers such as Martin Margiela began deconstructing clothing, exposing seams and linings, transforming deconstruction into an aesthetic language. A few years later, deconstructed minimalism became mainstream, absorbed and commodified by global brands.
More puzzling is that independent designers often see their creations reproduced by fast-fashion giants, even just a few weeks after a show. Zara and H&M have faced repeated accusations of copying the work of emerging designers to replicate it on an industrial scale.
Even more controversial are the cases of cultural appropriation. The influence of traditional motifs from exotic cultures, such as Mexican embroidery, indigenous beads, and Maasai patterns, to name a few, is part of a common way of operating, not just in fashion but in many other creative industries.
When copying becomes litigation
Fashion is the field where imitation often becomes the subject of legal litigation. The examples are numerous.
In 2009, Gucci America, Inc. sued Guess?, Inc. for trademark infringement, alleging that Guess had systematically copied the G logo, Gucci's green-red-green and green stripe, Gucci's signature braiding and false designation of origin. After years of global litigation, U.S. courts have awarded Gucci about $4.7 million and a permanent injunction against some of Guess's disputed designs; other jurisdictions have produced mixed results. The dispute was eventually resolved in 2018 on confidential terms, but it remains a milestone in fashion intellectual property litigation.
In the mid-2000s, Forever 21 faced more than 20 lawsuits from designers such as Diane von Furstenberg, Anna Sui, and others, who accused the company of copying their designs. Although few litigations resulted in final court verdicts due to legal hurdles in protecting clothing designs, some agreements were reached, and Forever 21 was permanently banned from duplicating specific DVF designs.
The fast-fashion giant Shein has been repeatedly sued by independent designers who accuse it of slavishly copying their registered designs. In some calls, the plaintiffs even argued that it was an AI-assisted replication of their original work. In January 2024, Uniqlo filed a lawsuit in Japan against Shein Japan for imitating its bestselling Mary Poppins shoulder bag.
In Star Athletica, LLC v. Varsity Brands, Inc. (2017), the U.S. Supreme Court ruled that aesthetic elements of clothing (e.g., graphic design patterns) can be copyrighted if they can be mentally separated from the garment's utilitarian function. This ruling represented a breakthrough in fashion copyright, expanding the legal tools designers can use to protect expressive design elements and challenging the age-old assumption that clothing itself could not be copyrighted.
The history of fashion emphasises a crucial point: copying in cultural production is neither new nor linear. When imitation becomes a total replica at industrial speed, including through the use of AI generation based on huge datasets of images, the stakes become ethical, legal, and philosophical.
Artificial Brain vs Human Brain
Artificial intelligence does not apprentice under just one or a few masters. He trains on vast datasets comprising millions, sometimes billions, of images and texts. He does not "remember" in human terms. Encode statistical relationships in latent space.
Yet, for the human artist whose aesthetic signature appears in the generated productions, the distinction between statistical coding and imitation may seem vain.
Humans copy selectively and slowly. AI synthesises collectively and instantaneously.
The drawback of copying is not new, as we have seen. In the case of AI, it is super accelerated.
We have always borrowed from the past. But now lending is automated, invisible, and amplified beyond human comprehension. This fact raises pressing questions regarding the authorship of generative art, creative ownership in the age of AI, and the evolving boundaries of intellectual property in machine learning systems.
Beneath the legal and economic debates lies a deeper philosophical inquiry. In fact, one wonders, when AI is the creative medium, what exactly does art become?
Prompt vs. Product
Sol LeWitt, a pioneer of conceptual art, wrote in 1967:
"The idea becomes a machine that creates art."
LeWitt often provided written instructions for wall drawings made by others. The physical design could vary, but the concept of the education was the real work of art.
Sound familiar?
In generative art, it is the prompt that functions like LeWitt's instruction. The AI model executes. The artist takes care of the outcome.
Yet there is a difference. LeWitt's assistants were human and visible. AI systems are trained on multitudes of invisible data.
Artists no longer directly manipulate pigments or other materials. Instead, they manipulate language, parameters, and constraints. The gesture moves upstream, from material action to immaterial instruction.
So, what is the artwork?
You could say that the prompt is the true artistic act. It is there that the vision crystallises. The codification of taste, the direction, the aesthetic judgment. The result is simply a materialisation of the instructions received, rendered by an algorithmic assistant. We could say that the AI is the brush, the dataset is the painting, and the prompt is the hand.
But prompts are often ephemeral. They can also be very short, functional, and iterative. Can we consider them expressive artefacts in themselves? Rarely. A prompt can be ingenious, but its value lies in what it produces.
Alternatively, we could say that the artwork is the result, such as an image, video, composition, music, infographic, or text. After all, that's exactly what the audience perceives. In the end, art has always been a sensory experience.
However, this alternative risks flattening authorship. If the result of iteration is the work of art, who is the artist? The executor of the prompt? The AI or LLM model? The engineers who trained him? The many creators whose works brought the dataset to life?
If this were the case, the artefact would become ontologically unstable.
We probably need a third, broader and more inclusive option.
The work, in this sense, could be the dialogue between prompt and result, that is, the iterative negotiation between human intention and the generation of machines.
In this vision, authorship becomes relational as the human provides direction, the artificial model offers variations, curation of the human becomes central, and selection becomes authorship.
The act of choosing, sifting, and rejecting hundreds of works and selecting one more closely resembles the art of the photographer than that of the painter. The camera captures, the photographer frames and decides.
Generative artists do not create pixels but curate spaces of possibility.
Conclusion
In light of the above, art, therefore, can reside not only in the prompt or in the product, but in the dynamic choreography between them. It can be defined by the consciousness that navigates between intention and system, by the ethical clarity that frames influence, and by contextual intelligence that transforms synthesis into meaning.
The artificial brain does not threaten art because it copies. It disturbs us because it makes visible how interdependent creation has always been.
But are we ready to redefine authorship itself?
Perhaps this is not the end of art, but the end of a certain illusion of art itself.
When Japanese woodblock prints reached Europe in the 19th century, they introduced a radical lesson. In fact, they were not unique masterpieces. They were reproducible prints. They could be mass-produced and therefore lacked the aura traditionally associated with the unique object.
Yet, they transformed Western art.
Hokusai's Great Wave of Kanagawa did not detract from painting. It destabilised painting. It influenced Impressionists, Post-Impressionists, designers, illustrators, photographers, and filmmakers. The East did not threaten European art, but expanded it.
The serial image, although devoid of the charm of uniqueness, demonstrated that aesthetic strength does not depend on singularity. It can be multiplied, shared, and distributed, while retaining its soul.
Western art did not collapse under reproducibility. Art reinvented itself.
Perhaps generative AI represents a similar wave.
The end of artistic innocence does not mean that art is dying. It means that we are finally recognising that art has always been steeped in influences, exchanges, even copying and transformation.
It was that wave that once swept over Europe from Kanagawa that made us discover that beauty could multiply without dissolving. That influence must not erase the original meaning of the artist. That replicability must not extinguish the spirit.
The real question is whether, in this Algorithmic Wave, we will allow large-scale production to bury creation, or if we will invent a way to allow authorship, recognition, and creative property to evolve.
Art has survived mass printing; it may even survive generative intelligence.
But what will be the fate of the creatives who work in its shadow?
Article: Astrea Nicodemo
Translation: Astrea Nicodemo
Images & Video: Eretikos Art
People also ask
How is authorship defined in AI-generated art?
Authorship in AI-generated art is increasingly understood as relational rather than singular. In traditional art, the author is the individual who executes the work. In generative AI, authorship may involve the person who designs the prompt, the system that generates the output, and the broader dataset that informs the model. Most legal and cultural frameworks currently attribute authorship to the human who directs and curates the process. However, the debate remains open, as AI challenges the idea of authorship as purely manual creation and shifts it toward conceptual orchestration and intentional selection.
Is AI art stealing from artists?
AI art is not automatically theft, but it raises legitimate concerns about how training data is sourced and used. Generative models learn from vast collections of existing images and texts, which may include works by living artists. While AI systems do not copy in a literal sense, they can reproduce stylistic patterns that resemble specific creators. The ethical and legal debate centers on consent, attribution, and economic impact. Whether AI art constitutes “stealing” depends on how datasets are built, regulated, and compensated.
Who owns the copyright of AI-generated art?
In most jurisdictions today, copyright protection requires human authorship. This means that fully autonomous AI-generated works may not qualify for copyright unless there is meaningful human creative input. Courts and copyright offices are still defining what constitutes “sufficient human contribution.” If a person meaningfully shapes prompts, selects outputs, and curates the final result, ownership may be attributed to them. However, unresolved questions remain regarding training data, derivative influence, and potential infringement claims. Copyright law is adapting, but it has not yet fully caught up with generative AI.
Is AI art truly original?
AI art is original in the sense that it generates new combinations of learned patterns, but it is not original in isolation. Generative systems do not create from nothing; they synthesize from vast datasets of existing human work. This mirrors how human creativity has always functioned, through influence, memory, and transformation. The key distinction lies in scale and speed. AI compresses centuries of aesthetic influence into seconds of generation. Whether this diminishes originality or redefines it depends less on the machine and more on how we understand creative transformation itself.
What are the ethical concerns surrounding AI in creative industries?
The main ethical concerns involve consent, compensation, attribution, and power asymmetry. Many generative AI systems are trained on large datasets that may include works by living artists who did not explicitly authorize their inclusion. This raises questions about fair use, economic impact, and recognition. There is also concern that AI may amplify existing inequalities in creative industries by scaling visibility for some while obscuring others. Ethical AI in art requires transparency in data sourcing, fair economic models, and frameworks that protect emerging creators from systemic invisibility.
How are artists responding to AI-generated art?
Artists are responding to generative AI in diverse and often contradictory ways. Some embrace it as a collaborative tool, integrating prompt engineering and algorithmic systems into their creative practice. Others critique it, highlighting concerns around labor, authorship, and appropriation. Many are experimenting with hybrid models in which human intuition and machine generation coexist. Rather than replacing artists, AI is prompting a redefinition of artistic roles, from fabricator to curator, from executor to orchestrator of possibility spaces.
Will AI change the future of art?
AI is likely to change how art is produced, distributed, and legally defined, but not the fundamental human impulse to create. Historically, technological shifts, from printmaking to photography, have disrupted artistic practice without eliminating it. Generative AI may represent a similar turning point. It challenges the myth of artistic innocence and forces reconsideration of originality, ownership, and influence. The future of art will not depend on whether AI can generate images, but on how societies structure power, recognition, and creative opportunity in response to that capability.
© 2026 FutureScape, a project by Astrea Nicodemo. All rights reserved. All original text, images, and video content published herein are protected by copyright. Unauthorized reproduction is prohibited.
Some visual materials may be created or enhanced using artificial intelligence tools under human creative direction.
Comments