The dealer and deepfakes: digital forgery and unauthorised reproductions

01The nature of the phenomenon: from artisanal copying to algorithmic forgery

Counterfeiting in the art world is not a new phenomenon. History is replete with famous forgers, from Han van Meegeren who deceived experts with his fake Vermeers in the 1930s to Wolfgang Beltracchi whose imitations of German Expressionists circulated for decades in the most prestigious collections. But these forgers worked with artisanal means: paint, canvas, aged pigments, artificial patina. Their production was slow, expensive and physically traceable. The artistic deepfake operates in a radically different register. An algorithm trained on an artist's corpus can generate in minutes hundreds of images that reproduce their motifs, palette, composition and touch with a precision that the non-expert eye cannot distinguish from an authentic work.

This reproductive capacity primarily affects two-dimensional works — paintings, drawings, photographs, prints — that naturally lend themselves to digitisation and algorithmic manipulation. But three-dimensional modelling technologies also allow the reproduction of sculptures, installations and art objects in digital files usable for 3D printing. The dealer representing a sculptor whose forms are recognisable must consider the possibility that unauthorised reproductions circulate as physical objects produced from digital scans of their works.

02Concrete risks for the gallery and its artists

The first risk deepfakes pose to the gallery concerns the market value of the works it represents. Scarcity is one of the foundations of an artwork's value: a collector agrees to pay a high price because they know the work they are acquiring is unique or produced in limited edition. If digital reproductions of comparable quality circulate freely online, this perception of scarcity is diluted, and with it the collector's willingness to invest. Galerie Gagosian has confronted the circulation on social media of AI-generated images imitating the style of some of its most recognised artists, leading to confusion among enthusiasts and novice collectors.

The second risk is reputational. A gallery whose artists fall victim to widely disseminated unauthorised reproductions may see its credibility questioned if it does not respond firmly and swiftly. Collectors expect their dealer to guarantee the authenticity of the works offered and to protect the integrity of their artists' output. A dealer who appears powerless against the proliferation of digital fakes sends a signal of vulnerability that can erode the trust of the most demanding clients.

The third risk concerns the artists themselves. An artist whose style is systematically copied by algorithms may experience a form of creative dispossession that affects their motivation and relationship with the market. Artist Trevor Paglen, whose work interrogates precisely the technologies of surveillance and artificial intelligence, has emphasised that the algorithmic reproduction of an artist's style raises ethical questions that exceed the existing legal framework. The dealer, as the artist's primary commercial interlocutor, must be able to understand these issues and support their artists in defending creative integrity.

03The legal framework: existing protections and gaps

Intellectual property law offers partial protections against unauthorised reproduction of artworks, but these protections collide with the technical reality of deepfakes. Copyright protects the original work, its reproduction and dissemination, but it does not protect an artist's style as such. An algorithm generating an image in the style of Gerhard Richter without reproducing any specific Richter work does not, under current European law, infringe the artist's copyright. This fundamental gap lies at the heart of the legal debate that the spread of generative artificial intelligence has opened in European and American legislative bodies.

The European AI regulation, adopted in 2024, imposes transparency obligations on providers of generative AI systems, including the obligation to indicate that content has been machine-generated. But this labelling requirement does not resolve the underlying problem: an image marked as AI-generated but faithfully imitating a living artist's style continues to circulate and be shared, and the marking is easily removed by a malicious user.

The dealer can nevertheless mobilise several legal tools to protect their artists. Trademark law, when the artist's name is registered as a trademark, offers protection against unauthorised commercial use of that name. The right to one's image, in jurisdictions that recognise it, may be invoked when AI generates portraits of artists in misleading contexts. Unfair competition and commercial parasitism offer avenues of recourse when a third party exploits an artist's reputation to market unauthorised derivative products. Galerie Pace maintains permanent legal monitoring to identify and halt unauthorised uses of its artists' works and names.

04Technological protection tools

Facing a technological threat, the response must partly be technological. Several solutions are emerging to help galleries and artists protect their works against unauthorised reproduction. Invisible watermarking technologies allow the embedding within a digital image of a unique identifier, imperceptible to the eye but detectable by specialised software, that traces the image's origin and proves it was extracted from an authorised source. These technologies, developed by firms specialising in intellectual property protection, are increasingly accessible to galleries of all sizes.

Provenance registries based on blockchain technology offer a complementary path. By recording each work and each transaction in a distributed, tamper-proof ledger, blockchain creates a verifiable ownership history that makes it harder to confuse an authentic work with a digital copy. Galerie König in Berlin has explored the use of blockchain-backed digital certificates to authenticate its artists' works in the digital environment.

Deepfake detection tools, initially developed to combat political disinformation, are beginning to be adapted to the art market context. These tools analyse the technical characteristics of an image — metadata, compression artefacts, statistical regularities revealing algorithmic generation — to determine whether it was produced by a human or a machine. Their reliability is improving rapidly, even though the race between generation and detection capabilities remains open.

05The dealer's responsibility in raising awareness

The dealer occupies a strategic position in the art market value chain for raising stakeholder awareness about deepfake risks. Towards collectors, they must emphasise the importance of acquiring from verified sources and warn against overly attractive offers circulating on non-specialist platforms. Towards artists, they must encourage the adoption of digital protection practices and ensure that high-resolution images of works do not circulate without precaution.

Galerie Thaddaeus Ropac has integrated into its practices a strict policy for managing digital images: high-resolution photographs of works are transmitted only to authorised institutions and publications, accompanied by complete metadata and explicit usage restrictions. This discipline, which may seem constraining, constitutes a first line of defence against the uncontrolled circulation of images exploitable by generative algorithms.

The dealer must also be able to explain to collectors the fundamental differences between an original work and a digital reproduction, however high its quality. The materiality of the work — the texture of the canvas, the thickness of paint, the traces of the artist's gesture, the patina of time — remains an authenticity criterion that digital reproduction cannot falsify. This physical dimension, paradoxically reinforced by the virtual threat, constitutes a selling argument the astute dealer knows how to deploy.

06Turning the threat into a value argument

The proliferation of deepfakes and unauthorised reproductions, concerning as it is, can paradoxically reinforce the value of the original work and the dealer's role as guarantor of authenticity. In a world saturated with machine-generated images, the certainty of owning a work created by an artist's hand, accompanied by a verified provenance certificate and a direct relationship with the representing gallery, acquires enhanced value. The dealer who articulates this value proposition to collectors transforms a threat into a competitive advantage.

Galerie Marian Goodman, whose artists are among the most exposed to unauthorised reproductions due to their international renown, maintains a rigorous authentication policy comprising catalogues raisonnés, detailed certificates and provenance tracking for every work. This rigour, far from being perceived as bureaucratic by collectors, is valued as a mark of seriousness and investment protection.

07Preparing the gallery for tomorrow's challenges

Generative artificial intelligence technologies continue to advance at a pace that legal frameworks and professional practices struggle to match. The dealer who anticipates these developments by training, investing in protection tools and cultivating a culture of digital vigilance within their team will be better equipped to face the challenges ahead. Training gallery staff to identify suspicious images, implementing systematic verification procedures for works acquired on the secondary market, and collaborating with cybersecurity experts are measures the forward-thinking dealer can adopt today.

For Artedusa partner galleries, the platform offers a secure environment where each presented work benefits from verifiable traceability and a direct association with its representing gallery. This transparency constitutes a guarantee for collectors wishing to acquire works with full confidence, in a context where distinguishing the authentic from the falsified is becoming a central concern of the contemporary art market.