In the sixteenth century, Andrea del Sarto painted a copy of Raphael’s portrait of Pope Leo X. According to Giorgio Vasari, the reproduction was so convincing that it fooled Giulio Romano, who had worked alongside Raphael himself. Andrea had copied everything, right down to the marks and blemishes on the original painting. When Romano eventually discovered the deception, he was reportedly even more impressed.
It’s a wonderful story, partly because it describes a relationship with copying that feels strangely foreign today. The ability to reproduce another artist’s manner was once considered a remarkable achievement, worthy of admiration regardless of who had invented the composition.
Mannerism emerged, in part, from this almost obsessive familiarity with the work of earlier masters. Following Michelangelo and Raphael, a generation of artists found itself working in the shadow of models that seemed to have already achieved an impossible degree of perfection. The answer was to look again, study harder, exaggerate, repeat. Painting increasingly began to feed on painting.
There is something remarkably contemporary about this attitude.
Five centuries later, much of visual culture still operates through the same mechanism. Images refer to other images, styles revive earlier styles, and artists reproduce the work of other artists. What has changed, perhaps, is the language used to justify it.
In the 1960s, Elaine Sturtevant began remaking works by contemporary artists, including Andy Warhol and Jasper Johns. Her repetitions appeared while the originals were still relatively new, sometimes before they had acquired their full historical significance. This detail matters. Sturtevant wasn’t simply revisiting art history. She was interfering with something still in progress, deliberately occupying another artist’s territory.
In 1981, Sherrie Levine photographed reproductions of Walker Evans’s celebrated photographs and presented them as her own works. The title, After Walker Evans, openly acknowledged the source. The gesture was almost irritating in its simplicity. Yet it worked precisely because we knew who Walker Evans was, because his photographs already carried cultural authority, and because we had been taught to consider authorship an essential part of an image’s value.
Without Walker Evans, Levine’s work would lose much of its tension.
That dependency, often treated as a conceptual problem, might actually explain part of appropriation art’s appeal.
Recognition is pleasurable. Encountering something familiar gives us the reassuring sense that we already possess some of the information required to understand what we are looking at. A reference flatters the viewer. It rewards cultural memory.
For an artist, borrowing a recognizable image means beginning with a relationship that already exists. The original has done much of the difficult work. It has entered the collective imagination, accumulated associations, acquired a certain authority.
How much of a copy’s success, then, belongs to the artist who produces it? And how much is inherited?
Richard Prince’s New Portraits series brought this ambiguity into particularly sharp focus. In 2014, Prince reproduced images taken from Instagram, including comments he had added beneath them. Some of the resulting works sold for around $90,000. The series provoked controversy, particularly among the photographers and subjects whose images had been used.
In response, the SuicideGirls collective offered prints of some of those same images for $90.
Two remarkably similar images. Two radically different prices.
At that point, the question of copying almost becomes secondary. Clearly, the market was purchasing something beyond the image itself. It was purchasing Richard Prince, his position within the art world, and the significance attached to his decision to reproduce it.
The artist’s name had become more valuable than the image’s origin.
This is where quotation can become something of a shortcut. A familiar image provides immediate recognition. Institutional context supplies legitimacy. The work can then be surrounded by a discourse about authorship, reproduction, ownership, or the instability of the original. By now, it is a vocabulary we know rather well.
Of course, some acts of appropriation fundamentally change how we understand an image. Others achieve little beyond reminding us of another artwork, preferably a more famous one.
The distinction isn’t always obvious.
Perhaps part of the difficulty comes from our tendency to locate originality in appearance. Two visually identical images can carry entirely different meanings depending on who presents them, when, and under what circumstances. But the observation that context changes meaning should not, by itself, make every repetition interesting.
Mannerism had already discovered the possibilities of working within an inherited visual language. Its most compelling achievements, however, did more than acknowledge the masters. Proportions became unnatural, bodies elongated, compositions grew unstable. Familiarity with the source made these deviations all the more striking.
Today, the opposite sometimes seems to happen. A recognizable image is retrieved precisely to preserve its recognizability. The visual distance from the original becomes almost negligible, while the explanation surrounding the work grows increasingly elaborate.
It is an economy of attention the art world understood long before social media.
A copy can benefit from the success of an original without having to repeat the process through which that success was achieved. It can borrow its authority, exploit its familiarity, and convert an existing cultural relationship into a new object of desire.
There is nothing inherently wrong with that. Art has always developed through imitation, influence, and the circulation of forms. The question is whether the repetition actually does anything with what it inherits.
And perhaps that leaves us with a rather simple, uncomfortable thought.
If we removed the original from the room, or even just erased our memory of it, how much of the work in front of us would remain?
There is a rather inconvenient way to test this idea
We could begin with one of the most influential artworks of the twentieth century, whose original has been missing for more than a hundred years.
Marcel Duchamp’s Fountain disappeared shortly after its controversial appearance in 1917. The urinal that helped redefine the boundaries of art is no longer available for anyone to examine. What survives are photographs, historical accounts, and a series of replicas produced or authorized by Duchamp in subsequent decades.
The version held by the Philadelphia Museum of Art dates from 1950. Another, in the collection of the San Francisco Museum of Modern Art, was fabricated in 1964. Neither is the object Duchamp originally submitted to the Society of Independent Artists. Nevertheless, both are presented as works by Duchamp.
The distinction is hardly trivial. A replica authorized by the artist becomes part of the legitimate history of the work. An identical object manufactured without that authorization would occupy an entirely different category.
Somewhere along the way, authenticity has become transferable.
The history of Fountain is particularly revealing because the original object was already an industrial product. Duchamp had appropriated something manufactured by someone else. When the original disappeared, another manufactured object could take its place, provided the substitution was approved by the artist.
The replacement was authentic because the right person had agreed to it.
This seemingly minor administrative detail has enormous consequences. It suggests that originality can survive the disappearance of the original object, while authenticity depends on a chain of decisions that may have little to do with its physical properties.
Sol LeWitt pushed this arrangement further. Many of his wall drawings were conceived as sets of instructions to be executed by other people. The Philadelphia Museum of Art acquired Wall Drawing No. 351 in 1982, purchasing what was essentially the right to realize the work according to the artist’s specifications. The drawing could be painted over and subsequently recreated.
The people physically making the drawing were not necessarily creating a copy in the conventional sense. They were producing an authorized realization of the artwork.
It would be possible for someone else to follow similar instructions and achieve an almost indistinguishable result. Yet the unauthorized version would not carry the same identity.
The interesting question is what, precisely, has been purchased.
A particular arrangement of lines and colours? A procedure? A contractual relationship with an artist? Or the exclusive privilege of presenting a certain configuration as an authentic LeWitt?
The answer involves all of these things, although their relative importance changes depending on whether one is speaking to an artist, a museum conservator, a curator or a collector.
This arrangement is not necessarily problematic. LeWitt’s practice established a rigorous distinction between conceiving an artwork and physically executing it. But it also exposes how much of contemporary art’s concept of authenticity has moved into the realm of documentation, ownership and institutional authorization.
And there is a curious contradiction here.
An authorized reproduction of a lost work can become an authentic artwork. Meanwhile, an entirely new painting, composed and executed by a skilled artist, can become a forgery simply because it bears the wrong name.
Wolfgang Beltracchi understood this particularly well.
For years, the German forger produced paintings in the styles of major twentieth-century artists, including Max Ernst, Heinrich Campendonk and André Derain. Many were not straightforward copies of existing paintings. They were invented compositions, presented as previously unknown works by established artists.
Beltracchi was effectively expanding the catalogues of artists who had never painted those pictures.
His operation required considerably more than technical imitation. The paintings needed plausible dates, convincing materials and credible histories. Together with his accomplices, he constructed false provenances that allowed the works to enter galleries, auction houses and private collections.
One painting, Red Picture with Horses, attributed to Heinrich Campendonk and dated 1914, sold for approximately €2.8 million in 2006. The deception was eventually exposed through scientific analysis. Traces of titanium white, incompatible with the painting’s supposed date, revealed that the attribution could not be correct.
The discovery changed nothing visible on the canvas.
Its composition remained intact. The colours had not deteriorated. The brushwork was exactly as convincing as it had been when experts and collectors accepted the work.
What collapsed was its history.
A laboratory result had transformed an expensive painting by Heinrich Campendonk into a fraudulent painting by Wolfgang Beltracchi.
There is something fascinating about how completely a work’s identity can be altered without making the slightest physical intervention to it.
It also raises an uncomfortable question about aesthetic judgement. If specialists genuinely admired these paintings before discovering their origins, what happened to that admiration afterward? Did they suddenly find the compositions less interesting, or had the wrong name made them impossible to appreciate in the same way?
The Beltracchi affair is not a defence of forgery. The deception caused genuine financial harm, and the fabrication of provenance was essential to the fraud. But the scandal demonstrates how easily the artistic assessment of an object becomes inseparable from the historical narrative attached to it.
The painting was original as a physical invention, false as a historical claim, and commercially valuable only for as long as the deception remained convincing.
Perhaps this is where the question of copying becomes genuinely interesting. Not in the similarity between two objects, but in the unstable relationship between an object and everything we are told about it.
Outside the commercial art world, however, the same problem takes on an unexpectedly different character.
In 1963, the prehistoric caves of Lascaux in France were closed to the public after years of tourism had damaged their delicate environment. The extraordinary paintings that had survived for millennia were being threatened by the presence of the people who wanted to see them.
The solution was to make copies.
Lascaux II opened in 1983, reproducing two of the cave’s most celebrated galleries. In 2016, Lascaux IV introduced a much more extensive facsimile of the original cave.
Millions of visitors have consequently experienced one of humanity’s most important artistic achievements through reproductions.
The original has effectively become too valuable to visit.
A similar approach was adopted for Tutankhamun’s burial chamber in Egypt. In 2014, a high-resolution facsimile produced by Factum Arte opened near the Valley of the Kings. The project was designed to reduce the damage caused by tourism while preserving public access to the experience of the ancient space.
These reproductions occupy an unusual position. Their purpose is to imitate as faithfully as possible, yet the ambition behind them is neither deception nor commercial appropriation. The better the copy performs, the greater its contribution to protecting the original.
In certain circumstances, the reproduction becomes the more responsible object to encounter.
This creates a problem for the traditional hierarchy in which originality automatically confers superiority.
A visitor entering a carefully reconstructed prehistoric cave may encounter spatial relationships, surface textures and painted figures with remarkable fidelity. The absence of the original material is an undeniable historical difference, but it does not necessarily prevent a meaningful aesthetic experience.
In fact, the facsimile can sometimes provide forms
THE RIGHT TO BORROW
Who gets to copy, who gets copied, and who gets paid.
In 1962, an illustrator named Tony Abruzzo drew a distressed young woman for Secret Hearts, a romance comic published by DC Comics. She was drowning, both literally and emotionally, with her boyfriend still visible in the background.
The following year, Roy Lichtenstein turned a detail from that illustration into Drowning Girl, one of the most celebrated paintings of American Pop Art.
Lichtenstein removed the boyfriend, tightened the composition, altered the dialogue and enlarged the image to the scale of a painting. What had been one moment in a disposable comic-book narrative became an autonomous object, eventually entering the collection of the Museum of Modern Art in New York.
Abruzzo’s original drawing remained a comic-book illustration.
There is something worth investigating in that movement between categories. The visual material had not changed beyond recognition. Its cultural status, however, had changed enormously.
Lichtenstein was interested in the visual conventions of mass reproduction, particularly the mechanical appearance of printed comics. His paintings introduced those conventions into a medium traditionally associated with individual expression and artistic prestige. The result was an intelligent disruption of the boundaries between commercial and fine art.
But there was another transformation taking place, less frequently discussed during the artist’s rise to fame. The creative labour of illustrators was being absorbed into works attributed to someone else.
The original comic artists did not receive comparable recognition, and their names were generally absent from the public narratives surrounding Lichtenstein’s paintings.
It would be simplistic to reduce Lichtenstein’s achievements to plagiarism. His work involved deliberate decisions about composition, scale, colour and the relationship between painting and commercial printing. But acknowledging those decisions does not eliminate the contribution of the artists whose images made them possible.
The issue becomes particularly interesting when we consider the hierarchy of the two professions.
An illustrator working for a comic-book publisher was producing commercial material within an industrial system. A painter working with that same material was producing cultural criticism. The distinction was accepted so readily that the second activity appeared intellectually superior almost by definition.
The painting was celebrated for questioning the status of the image. The person who had first created the image was rarely included in the conversation.
This is one of appropriation’s more revealing asymmetries. The freedom to borrow has never been distributed equally across the cultural landscape.
In 1988, Jeff Koons produced String of Puppies, a painted wooden sculpture depicting a seated couple holding eight puppies.
The sculpture closely resembled a photograph by Art Rogers, who had created the image several years earlier and sold it as a postcard. Koons had purchased a copy and used it as a reference for the sculpture.
Rogers sued.
Koons argued that his work was a parody of sentimental popular imagery, part of a broader examination of kitsch and consumer culture. In 1992, a federal appeals court rejected his fair-use defence, finding that the sculpture had copied protected expressive elements of Rogers’s photograph.
The dispute is interesting because Koons had undoubtedly changed the medium. A small black-and-white photograph had become a three-dimensional, brightly coloured sculpture. The transition was substantial in physical terms.
It was not enough.
The court examined the relationship between the two works rather than simply accepting the artist’s stated intentions. Recreating a photograph as a sculpture did not automatically make the borrowing permissible.
Fourteen years later, Koons found himself on the opposite side of a similar legal argument.
His painting Niagara incorporated a portion of a fashion photograph by Andrea Blanch, originally published in Allure magazine. This time, the court accepted the fair-use defence. The borrowed image had been repositioned within a larger composition and used for a sufficiently different purpose.
The two decisions are instructive when considered together.
There is no simple percentage of visual alteration that converts copying into legitimate artistic expression. A change of medium can be insufficient. A relatively modest use of recognizable photographic material can be acceptable.
What matters, at least under the American fair-use framework, is a combination of circumstances: the purpose of the use, its commercial character, the nature of the original material, the extent of the borrowing and its effect on the potential market.
These are legal considerations, of course, rather than universal criteria for artistic quality.
A work can be legally transformative and aesthetically unremarkable. It can also be artistically extraordinary while infringing someone else’s copyright.
The art world and the courts are not necessarily trying to answer the same question.
Yet their decisions have become increasingly entangled, particularly as appropriation has moved from an experimental artistic strategy into an established commercial practice.
One of the most consequential examples concerns Andy Warhol, a photographer named Lynn Goldsmith and a portrait of Prince.
In 1981, Goldsmith photographed the musician, then still approaching the height of his international fame. Three years later, Vanity Fair licensed one of her images for $400 so that Warhol could use it as a reference for an illustration.
Warhol produced a portrait and developed additional versions using the same source photograph.
In 2016, following Prince’s death, the Andy Warhol Foundation licensed one of those images, Orange Prince, to Condé Nast for $10,000. Goldsmith received nothing from that later transaction.
The legal dispute eventually reached the United States Supreme Court.
In May 2023, the Court ruled that the specific commercial licensing of Orange Prince did not favour the Foundation’s fair-use defence under the first statutory factor. The Court emphasized that Goldsmith’s photograph and the licensed Warhol image were being used for substantially similar purposes: illustrating magazine coverage of Prince.
The ruling did not declare Warhol’s entire Prince Series unlawful. It addressed the particular licensing transaction, leaving other uses of the artworks outside the scope of its decision.
Still, the financial relationship is difficult to ignore.
Goldsmith created the photograph. Warhol transformed its appearance. Decades later, his foundation received substantially more for licensing a derivative image than Goldsmith had originally received for the photographic reference.
Warhol’s cultural importance is hardly in question. The difficulty lies in assuming that his importance somehow cancels the economic interests of the person whose work he used.
The case exposed an uncomfortable feature of contemporary appropriation: the artist’s conceptual intervention can become valuable enough to obscure the professional labour on which it depends.
For photographers, illustrators and commercial image-makers, that labour has never been purely conceptual. It is also how they earn a living.
The distinction might appear mundane, but it is precisely where the discussion becomes less comfortable for the art establishment.
A photographer’s licensing fee can be treated as a commercial concern. A celebrated artist’s resale value can be treated as evidence of cultural significance.
Both involve money. Only one is routinely presented as something more elevated.
There is another possibility worth considering. Perhaps certain forms of appropriation succeed because they reproduce an existing social hierarchy rather than challenge it.
For much of the twentieth century, commercial imagery occupied a lower cultural position than painting or sculpture. Advertising, fashion photography, illustration and graphic design were generally associated with practical functions, clients and deadlines.
Fine art enjoyed a different status. It was understood as a space in which images could be interpreted rather than merely consumed.
When artists borrowed from commercial culture, that movement was often celebrated as a critique of the distinction between high and low art.
But the traffic mostly flowed in one direction.
An advertisement could become art when an artist reproduced it. A comic panel could become a museum masterpiece when enlarged onto canvas. A fashion photograph could acquire conceptual depth when removed from a magazine and inserted into an exhibition.
The reverse transformation carried considerably less prestige.
A designer reproducing an artwork for an advertising campaign was unlikely to be praised for questioning the conventions of artistic originality. The same gesture could easily be dismissed as derivative, opportunistic or simply unlicensed.
The difference was partly contextual, but also institutional. Some professions were granted the privilege of interpreting culture; others were expected to supply its raw material.
Appropriation art was capable of revealing this hierarchy while simultaneously benefiting from it.
This doesn’t invalidate the work. It does, however, complicate the flattering story that borrowing automatically represents a rebellious or democratic gesture.
At a certain point, the ability to appropriate becomes another privilege attached to cultural authority.
The invisible first author
The internet has made these relationships more visible, largely because the original sources are becoming easier to locate.
An image that once circulated anonymously in a magazine archive can now be identified within seconds. Side-by-side comparisons reveal compositional similarities that might otherwise have remained buried in private collections or specialist publications.
The discovery does not necessarily diminish the later artwork. Sometimes it makes the artistic intervention more interesting. Sometimes it reveals how little was changed.
More importantly, it restores the presence of people who had almost disappeared from the story.
The illustrators, photographers and designers whose work supplied the source material are no longer entirely abstract figures. They have names, careers, professional circumstances and, in some cases, their own accounts of what happened.
This changes the nature of the debate.
For decades, appropriation was primarily discussed through the intentions of the appropriating artist. What was Warhol saying about celebrity? What did Lichtenstein reveal about mass media? What did Koons intend to expose about consumer desire?
Those questions remain relevant. They are simply incomplete.
The person whose work was borrowed might have understood the image very differently. The photographer may have been interested in lighting, expression and a particular encounter with the subject. The illustrator may have been working within the visual economy of a serialized narrative. Their contributions had meaning long before a celebrated artist discovered them.
There is no reason to assume that the second interpretation is automatically more sophisticated than the first.
And there is a peculiar irony in an artistic tradition that has spent decades dismantling the mythology of individual authorship while continuing to organize its own history around a relatively small number of exceptionally famous individuals.
We know the artists who appropriated. We are only beginning to recover the artists who were appropriated from.
The question is no longer simply whether copying can produce art. That argument has been settled by decades of artistic practice, critical writing and institutional recognition.
What remains unresolved is the distribution of credit within that process.
A source image can be altered, enlarged, recoloured, recontextualized and transformed into a successful artwork. Its original author can disappear almost completely, even while the image they created continues to circulate.
The new work may have genuinely changed the meaning of the old one.
It has also changed who gets remembered for making it.
THE ORIGINAL, IN STOCK
How contemporary art learned to manufacture recognition.
In January 2012, Damien Hirst occupied all eleven Gagosian galleries simultaneously. The exhibition, The Complete Spot Paintings 1986–2011, brought together more than 300 paintings borrowed from collectors and institutions across twenty countries. It was a spectacular logistical exercise, spread across eight cities, and an unusually effective demonstration of how far a single visual idea could travel. Visitors encountered circles of colour arranged in precise grids, repeated across canvases of radically different sizes. Some contained only a few spots; one contained 25,781. The paintings were recognizable almost immediately, and this was perhaps their most remarkable property. There was no need to identify a particular work. Knowing the series was enough. Gagosian even organized a challenge in which visitors who reached all eleven gallery locations could receive a personally dedicated Hirst print. The journey around the world had become another way of participating in the artist’s system of repetition.
By then, the production of Hirst’s spot paintings had long been delegated to studio assistants. Rachel Howard, who began working for him before his operation developed into a large-scale enterprise, became particularly accomplished at painting them. Hirst openly praised her technical ability, even describing the best of his spot paintings as one executed by Howard. The admission is interesting for reasons beyond the familiar debate about whether an artist should physically produce their own work. Howard was herself an artist, with an independent practice and her own understanding of painting. Yet when she painted Hirst’s circles, the resulting object belonged to his artistic identity. Her technical expertise was incorporated into a visual system that had already established its author. The work did not need Hirst’s hand to be recognized as a Hirst. It needed the rules, the distribution network and the signature.
This arrangement resembles the production of luxury goods more closely than the romantic image of an artist working alone in a studio. The designer establishes a vocabulary, the workshop ensures consistency, and the resulting objects circulate under a recognizable name. Consistency is essential because the buyer must be able to identify the product without examining its entire history. An unfamiliar Hirst might require an explanation; a familiar one can announce itself from across the room. Repetition makes that possible. It gradually turns a particular artistic decision into something resembling a trademark. There is, however, an important distinction between a recognizable style and a commercially managed identity. Artists have always developed recurring formal characteristics. What appears relatively new is the degree to which those characteristics can be organized into a production system, reproduced across categories and maintained as a coherent public image. The work becomes easier to recognize with every additional example, even when the individual example contributes very little to the development of the original idea.
Takashi Murakami has explored this territory with particular sophistication. His company, Kaikai Kiki, operates across art production, artist management, publishing, merchandising and other commercial activities. Murakami’s smiling flowers appear in paintings, sculptures, prints, clothing, accessories and digital projects, circulating through environments with very different expectations about price and artistic significance. In December 2024, Gagosian opened a presentation at Burlington Arcade in London that made the relationship unusually explicit. Four new flower paintings were displayed in the upstairs gallery, while the shop offered clothing, cushions, soft toys, prints and other merchandise featuring the artist’s imagery. The arrangement was almost too perfect as a demonstration of contemporary art’s expanding commercial vocabulary. Upstairs, the collectible artwork; downstairs, its distributed identities. Yet the familiar distinction between the authentic object and its cheaper commercial relatives seems inadequate here. The merchandise is not an embarrassing side effect of Murakami’s artistic practice. It belongs to the structure through which his images acquire visibility, circulation and cultural relevance.
The smiling flower is especially effective because it can survive nearly any change of material or scale. Printed on a handbag, reproduced as a cushion or painted on canvas, it remains unmistakably Murakami. Its identity is no longer tied to a particular physical manifestation. It resides in a set of visual characteristics that can be transferred almost indefinitely. This is familiar territory for commercial brands, which spend enormous sums trying to make their symbols instantly recognizable. In Murakami’s case, however, the same process belongs simultaneously to a serious artistic practice, one informed by Japanese visual history, consumer culture and the aesthetics of mass production. The multiplication of the image is not merely a business strategy added after the fact. It is part of the world the artist has chosen to construct. What deserves scrutiny is the ease with which contemporary audiences now move between these categories. We can encounter the same motif as a museum object, a luxury product and an inexpensive collectible, adjusting our expectations almost automatically according to where it appears.
KAWS offers an equally revealing example. His COMPANION character first appeared as a limited-edition vinyl figure produced with the Japanese brand Bounty Hunter in 1999. Its crossed-out eyes, skull-like head and familiar cartoon anatomy gave it an immediately identifiable appearance. Over the following decades, the character expanded into monumental sculptures, paintings, collectible objects and public installations. By the time the Brooklyn Museum presented KAWS: WHAT PARTY in 2021, the distinctions between toy, sculpture and popular icon had become central to understanding the artist’s international success. A small figure purchased as a collectible and a gigantic version installed in a public square could belong to the same visual family while operating in entirely different economic and institutional environments. The enlarged sculpture did not need to reinvent the character. Much of its appeal depended on viewers already knowing it in another form.
This is an important development in the history of artistic repetition. Traditionally, an artist might produce copies or variations after an original work had achieved recognition. Here, the repeated character is the original proposition. Its identity is constructed through continuous appearances rather than established by a single definitive object. One version does not necessarily occupy a privileged position above the others. Instead, the various objects strengthen each other’s visibility. A museum exhibition lends prestige to the collectible toy, while the widespread popularity of the toy ensures that audiences recognize the museum sculpture. Cultural value moves in both directions, although financial value remains carefully differentiated through size, materials, editions and distribution.
Limited editions become particularly useful within this arrangement. They allow an image to circulate widely while restricting the availability of specific objects. A figure can be produced in hundreds or thousands of units, each identical within its edition, while scarcity is maintained through production limits, colour variations and controlled releases. The industrial logic of reproduction is combined with the financial logic of exclusivity. This may sound contradictory, but the system is familiar from luxury fashion, sneakers and collectible design. Scarcity does not necessarily require a unique image. It requires a limited opportunity to possess a particular authorized version of that image. The difference is considerable. An image can become almost impossible to avoid in everyday culture while the objects carrying it remain expensive, difficult to obtain or highly desirable.
There is an unresolved tension between the two ambitions. Artists need visibility to develop recognizable identities, but excessive visibility can also diminish the sense of discovery on which artistic interest depends. A motif repeated too frequently risks becoming decorative shorthand for an artist’s reputation. The problem is difficult to measure because commercial success can continue long after the original artistic proposition has stopped developing. Indeed, the more familiar the motif becomes, the less formal experimentation may be required to maintain its public recognition. Collectors know what they are buying, galleries know what audiences will recognize, and manufacturers know which characteristics must remain unchanged. An aesthetic originally associated with experimentation can gradually become a system for avoiding it.
The most revealing aspect of this economy is the disappearance of any clear need for a first object. Someone encountering a Murakami flower or a KAWS figure today may have no knowledge of when the image was created, which version appeared first or how its visual language developed. None of this prevents recognition. The image has acquired an identity independent of its origin, circulating through so many formats that its history becomes less important than its continued presence. The audience recognizes the artist without necessarily knowing a particular artwork. In commercial terms, this is an extraordinary achievement. In artistic terms, it introduces a more complicated relationship between familiarity and understanding.
An image that can be reproduced indefinitely does not become meaningless simply because it is familiar. Repetition can deepen associations, alter cultural context and give forms a significance that a single appearance could never produce. But it can also preserve an artistic identity in a state of permanent recognition, allowing successive objects to inherit attention from an established visual formula. The commercial success of these objects may therefore tell us less about the invention contained in each individual work than about the efficiency of the system connecting them.
Perhaps the most unusual achievement of this system is that it has made copying unnecessary in the traditional sense. The artist no longer needs to reproduce somebody else’s successful image. Once their own image has acquired sufficient recognition, they can begin reproducing themselves. And the market, far from regarding each subsequent version as a dilution of the first, may understand the repetition as confirmation of its importance.
At that point, originality becomes an increasingly difficult quality to locate. The first version may have established the language, but it is the accumulation of later versions that makes the language valuable. The image’s authority appears to grow retrospectively, as though each new object were producing additional evidence that the original deserved recognition all along.
The contemporary art market has learned to sell this process with remarkable efficiency. It offers the pleasure of owning something familiar, accompanied by the assurance that the particular object remains sufficiently scarce.
One can now buy an original without ever encountering anything unexpected.
THE COPY WITHOUT AN ORIGINAL
What happens when we can reproduce something that never existed?
In 2016, a group of researchers and technologists unveiled a painting by Rembrandt. The Dutch artist had been dead for almost 350 years, but the portrait looked reasonably convincing: a middle-aged man in dark clothing, a black hat, the familiar handling of light across his face. The image was the result of a project called The Next Rembrandt, developed through a collaboration involving ING, Microsoft, Delft University of Technology and Dutch cultural institutions. Researchers had analysed hundreds of paintings, studying everything from facial proportions and compositional preferences to brushstroke patterns. The resulting image was printed using technology that recreated something resembling the texture of oil paint. The project attracted considerable attention, although the painting itself was perhaps less interesting than the problem it introduced. Nobody was claiming to have discovered a lost Rembrandt. Nobody was attempting to forge a particular canvas. The intention was to produce an entirely new work that looked as though Rembrandt might have painted it. The result was a copy of an artistic identity rather than a copy of an artwork.
This distinction becomes increasingly difficult to ignore. A traditional copy begins with an identifiable source, even when that source is subsequently disguised. Here, the reference was dispersed across an entire body of work. The system identified recurring characteristics and assembled them into a plausible new composition. It was effectively reproducing what we recognize as Rembrandt without reproducing any single Rembrandt painting. But what exactly makes a painting look like a Rembrandt? A particular arrangement of light, a restricted palette, the treatment of skin, the dramatic relationship between illuminated faces and dark backgrounds. These characteristics are familiar to anyone with a reasonable knowledge of seventeenth-century Dutch painting. The project demonstrated that familiarity could be converted into instructions. It also exposed how easily an artist’s identity can be reduced to the features most commonly associated with their work. Rembrandt’s more unusual decisions, his failures, his contradictions and the changes that occurred throughout his career become difficult to accommodate within a system designed to produce the most statistically recognizable version of his style.
Two years later, a rather unremarkable portrait of a fictional gentleman appeared at Christie’s in New York. Portrait of Edmond de Belamy, created by the French collective Obvious using a generative adversarial network, looked like a deteriorating eighteenth- or nineteenth-century painting. The man’s face was blurred, the composition slightly awkward, the entire image suspended somewhere between a historical portrait and a photographic accident. The algorithm had been trained on approximately 15,000 portraits spanning several centuries. The work sold for $432,500 in October 2018, vastly exceeding its estimate. Much of the press coverage focused on the extraordinary idea that a machine had created an artwork valuable enough to enter the auction market. Yet a less glamorous story was unfolding behind the scenes. The collective’s technical approach relied heavily on code and methods made publicly available by Robbie Barrat, a young programmer who had been experimenting with generative art. Barrat had shared his work openly, but he had not expected another group to turn a closely related process into a high-profile commercial success without initially acknowledging his contribution. The controversy was documented by Wired shortly after the sale.
The episode has the messy quality of a particularly contemporary art dispute. An algorithm trained on historical paintings produces an invented portrait. A collective claims artistic authorship over the result. The software owes a substantial debt to a programmer whose name is initially missing from the story. An auction house sells the object for almost half a million dollars, while much of the resulting attention celebrates the supposed autonomy of the machine. By the time the portrait reaches the buyer, authorship has become a surprisingly crowded affair. It is worth remembering that the algorithms themselves had not developed an interest in portraiture. Someone selected the material, someone wrote the code, someone chose the final image, and someone decided that printing it, framing it and placing it at auction constituted an artistic proposition. The machine made certain operations possible. The surrounding decisions remained conspicuously human.
What has changed since then is the accessibility of the technology. Creating an imitation of a historical style no longer requires a specialized research project or a custom-trained neural network. Generative image systems have made such exercises available to a vastly larger public. One can produce an imaginary Renaissance painting, a nonexistent fashion campaign from the 1990s or a photograph resembling a forgotten family archive. The results do not necessarily reproduce any particular image, although generative systems can sometimes reproduce recognizable elements from their training material. More commonly, they offer combinations of visual conventions that make the image appear to belong to a familiar category. A new picture can arrive already looking old, already carrying the surface characteristics of something we feel we have seen before.
In April 2023, the German artist Boris Eldagsen refused a prize he had won in the Creative category of the Sony World Photography Awards. His submission, Pseudomnesia: The Electrician, depicted two women in an unsettling image resembling an old photograph. The picture had been generated using artificial intelligence. Eldagsen had entered it to provoke a discussion about the boundaries of photography, although the organizers disputed aspects of his account of how its origins had been disclosed. His refusal became international news. The particularly interesting feature of the image, however, was its apparent historical character. It looked as though it had been taken decades earlier, under circumstances that the viewer could only imagine. The women seemed to possess a history. The grain, the tonal range and the imperfect details suggested the physical survival of a photograph from another period. None of that history existed.
This is perhaps a more significant development than the ability to imitate a famous painter. Photography has always carried a particular relationship to the past. Even when manipulated, staged or misleadingly presented, a conventional photograph begins with light registering something in front of a camera. Eldagsen’s image borrowed the appearance of that relationship without requiring the event itself. The photographic surface was convincing enough to suggest documentary evidence, although there had been no encounter between a photographer and those two women. They had never stood together for the picture. There was no forgotten negative, no family album, no moment that could be recovered through archival research. The image had acquired the appearance of memory without the inconvenience of an actual past.
This matters beyond the debate over whether artificial intelligence can produce art. We are becoming accustomed to images whose visual references are abundant but whose individual origins are difficult, sometimes impossible, to identify. In this environment, resemblance begins to operate differently. We may recognize the atmosphere of a 1970s magazine advertisement without being able to point to the advertisement itself. We may identify a vaguely cinematic composition, a familiar painterly treatment or a particular photographic mood, although the image has no direct historical counterpart. Its success lies in its ability to activate visual memory without making a specific citation. The source becomes an entire cultural archive, reduced to a set of recognizable tendencies.
There is considerable creative potential in this development. Artists can construct fictitious histories, invent convincing visual documents and explore forms of imitation that would previously have required enormous technical resources. But the ease of producing a recognizable style also changes its significance. When the appearance of a medium, period or artistic movement can be summoned almost immediately, its most familiar characteristics risk becoming a substitute for its actual complexity. The past turns into an aesthetic resource, conveniently stripped of the social circumstances, technical limitations and personal decisions that produced it. We recognize the texture of an era while knowing very little about the people who lived through it.
The same problem extends to the contemporary obsession with visual coherence. Images are increasingly expected to fit established categories, from editorial photography and luxury advertising to specific artistic movements. Generative systems are particularly efficient at satisfying these expectations because recognizable conventions are precisely what make a description operational. An instruction referring to a known style gives the system something to work with. The less familiar the intended result, the more difficult it may be to describe through the vocabulary of existing images. There is an irony in this arrangement. Tools associated with almost unlimited creative possibilities can encourage the production of images that remain remarkably close to what has already been seen.
None of this makes technological production inherently less artistic. A technically sophisticated oil painting can be empty, just as a generated image can be part of a rigorous and original investigation. The problem is not the absence of manual labour, although that is often where criticism begins. It is the possibility of producing the appearance of an artistic decision without having made the decision that originally gave the appearance its meaning. An image can acquire the visual symptoms of experimentation, age, damage or spontaneity without any of the processes that those characteristics once implied. What looks like an accident may be a requested effect. What resembles accumulated history may have been produced in seconds.
Returning to the original question of copying, the situation now appears considerably less stable than it did at the beginning. For centuries, imitation involved a recognizable relationship between an existing work and a subsequent one. Even when artists challenged the idea of originality, the older image remained somewhere in the background. It could be identified, compared, reinterpreted or disputed. Today, an image may derive its entire identity from things that resemble sources without ever having existed as individual originals. This makes the familiar argument about whether copying is legitimate increasingly inadequate. There may be no specific object to accuse the image of copying, while its visual language remains almost entirely inherited.
Perhaps the real scarcity in contemporary visual culture is no longer the ability to produce a unique image. Technology can generate endless variations, each technically distinct from the last. What remains difficult is producing an image that deserves to exist for reasons beyond its resemblance to something else. Recognition is a powerful advantage, and the art market has repeatedly demonstrated its willingness to reward it. But recognition can also become a rather comfortable form of artistic inertia, particularly when the pleasure of familiarity removes any expectation of discovery.
There is a detail in Vasari’s account of Andrea del Sarto’s copy of Raphael that now seems unexpectedly relevant. Andrea reproduced the stains and marks on the original painting because those imperfections were necessary to sustain the illusion. They belonged to the history of a real object, and copying them required observing that object with extraordinary attention. Five centuries later, we can manufacture the same signs of age, wear and authenticity without needing an original painting at all.
The stains can be invented too.
