Editor’s note: This story is the latest edition of Link Rot, a new column by Shanti Escalante-De Mattei that explores the intersections of art, technology, and the internet.
At the Palazzo Diedo, you can find a typically Venetian pleasure: seeing the sleek edge of contemporary art set against the backdrop of a fading fresco. And what an edge. “Strange Rules”—curated by Mat Dryhurst, Holly Herndon, and Hans Ulrich Obrist with Adriana Rispoli—gathers works of such strange futurity as fortune-telling metal plants, a film to be made over the course of a millennium by artificial intelligence, two-headed flatworms, and hyperscans of brains communicating in a pre-verbal state. In this show, the curators seem to be staking out a position on what counts as art, just as the mass adoption of blockchain, generative AI, and other disruptive technologies challenges our understanding of artistic authorship. Where is the line between content and performance art? Between scientific research and artistic research? Is using AI to make an artwork evidence of inexcusable laziness, or simply one more mechanical medium that, in the right hands, can be wielded to stimulating ends? Is a given website a gimmick, a game, or something to take seriously? These questions tend to boil down to one: “Is it art?”
It may be the most clichéd, surface-level inquiry one can make about contemporary art, and yet—to the frustration of artists who work with technology—it is a question they are made to answer again and again. With “Strange Rules,” the curators have devised a frame, and a shorthand, that makes the practices of a wide range of practitioners legible: “Protocol Art.”
An installation view of Ken Stanley’s Picbreeder (2008-2010) at “Strange Rules.”
Courtesy Berggruen Arts & Culture/Photo Joan Porcel
Protocol Art is, per the exhibition text, “a practice that engages with the underlying rules that determine how culture is produced, distributed, and perceived in a digital age. These rules manifest as algorithms, AI models, platforms, technological infrastructure and social convention, hardening through use into the conditions of cultural life. Protocol Art operates at the level of the rule: not only analyzing these systems but seeding new ones. The protocol is not the subject of art: it is the art. When the protocol is the art, the scientists and researchers authoring such rules become artists.”
The Palazzo Diedo—recently renovated and run by Berggruen Arts & Culture—means to become the temple of Protocol Art, and not only because of “Strange Rules.” More research and exhibitions on the practice are forthcoming.
At first blush, the definition might seem overly broad—like the same golden-ratio diagram laid over a picture of a cat, a supermodel’s face, a painting, and a sunset. The show makes a display of the wide range of work that can coexist under this model, from biologist Michael Levin’s Double-Headed Planaria—in which his research on, and manipulation of, genetic code is the relevant “protocol”—to Trevor Paglen’s Voyager (2026), a self-guided meditation in which language (protocol!) generates images in the mind of the audience member. But Protocol Art is perhaps most useful when we consider what it excludes: an illustrator painting concept art on his Wacom tablet, someone prompting a text-to-video model to make soap operas starring humanoid fruit, or anyone else using widely available technologies to create a visual output without investigating, intervening in, or hijacking its infrastructural and social logics. This is not to suggest that illustration and AI video content demand the same creative skills, only that Protocol Art makes swift work of distinguishing itself from work shaped by, let’s say, external tastes: those of the company commissioning the illustration, or the algorithmic pressures and analytics that mold the content posted by “creators.”
Protocol Art proposes another crucial distinction. If most people tend to think of the art as the visual or sonic manifestation one encounters, Protocol Art locates the art not in its manifestations but in the work that generates them. To give one example: digital artists can adjust the dimensions of a work as needed, so that it can be displayed on screens of multiple sizes and orientations. The “location” of the art isn’t fixed. By that logic, the authorial content of the work resides in the structure the artist has designed, not in any particular manifestation. This mode of authorship is key not only to grasping the malleability of an audience’s visual encounter with an artwork, but to understanding that the artistic encounter engendered by technologies like blockchain and AI is highly participatory and, often, specific to the viewer.
Works on view like Ken Stanley’s Picbreeder (2008/2020), Primavera De Filippi’s Plantoid (2014), and Dryhurst and Herndon’s Attention Guild (2026) feature a participatory element that guides the development of each piece’s form over time. Take Picbreeder, in which two sleek, glowing forms—one resembling a car, the other a frog—appear onscreen. Both images were generated as users navigated a website Stanley designed to let them pick their preferred variants of images, which then became the basis for a new set of generated images. Stanley, a researcher in artificial intelligence and machine learning, created a kind of precursor to the training later experienced by early users of generative AI models, in which a prompt returns four images, the user selects one, and four more are made. Over time, this selection process produced a kind of mass AI aesthetics, as the models learned average human preferences from millions of signals and gradual adjustments. These models, like the works around them, are impossible without a high degree of participation. But even though the final form of Stanley’s work would not exist without participation, that final form is not, per the curators’ understanding of Protocol Art, where the authorship lies. The authorship lies in the orchestration of participation.
Yet while Protocol Art is a neat and productive frame for understanding how the relationship between art and author is shifting in the modes of artmaking on view in “Strange Rules,” it is also a shield. The show and its framing build an argument about authorship that artists can point to when defending their use of certain tools, such as AI, as cultural institutions and legal frameworks mobilize some forms of resistance against them, irregardless of how they are being manipulated.

Fabien Giraud, The Final Epoch 1, 2025 – 2026.
Courtesy Berggruen Arts & Culture/Photograph David Levene
The existential drive for differentiation underlying Protocol Art is clearest in the case of Feral (Epoch 1) (2025–3025), a film shot and edited by AI and meant to unfold over the course of one thousand years, through the successive interventions of 32 generations of artists. Fabien Giraud is responsible for Epoch 1; Pierre Huyghe is tapped for Epoch 2; and after that, who knows—the artist may not even be born yet.
The team behind The Feral, as the overall project is known, has been at the forefront of the battle over AI authorship. Last September, Grégory Chatonsky and Ida Soulard, both members of the Feral team, published an op-ed in the French newspaper Libération taking issue with the National Center for Cinema and the Moving Image’s (CNC) decision not to finance films made with AI. Chatonsky, a co-scriptwriter for the AI, and Soulard, head of The Institute, the research and education platform that developed The Feral, argue that while the AI images and videos proliferating on social media might seem to vindicate the CNC’s decision, the real issue isn’t the technology but the absence of an author.
“What a machine without an author produces has a name: slop,” Chatonsky and Soulard write. That is to say: anyone using text-to-video generative AI is not an author.
But then what form of authorship is possible in a world “permeated by generative AI,” as Chatonsky and Soulard put it? Here is their definition: “The author chooses their models, trains their corpus, configures their processes, works locally with open-source software, and accepts the material consequences of their choices. Authorship no longer resides solely in the final image; it now permeates the entire operational chain.” Their definition aligns perfectly with the boundaries of Protocol Art.
Voids of authorship abound. Currently, the US Copyright Office does not recognize work made with AI as having an author. Though the effect is positive as it protects the livelihoods of commercial artists, the legal abyss it produces seems unbearably strange: where, then, does the work come from? For Chatonsky and Soulard, the ontological possibility of authorlessness is not incorrect, just too broadly applied.
Am I suggesting, then, that Protocol Art is a mere purification ritual, meant to separate the valid from the invalid, the artist from the teeming masses? No —though let’s leave the matter of film, AI, and funding to one side. Or maybe yes—but that’s not necessarily a bad thing. This separating of spheres feels particularly crucial given the embarrassment surrounding NFTs that still lingers among many art professionals. If there is one lesson that household names like Marcel Duchamp and Andy Warhol have taught the public, it is that “anything” can be art. By gesturing (metaphorically) toward the sign at the door of the MoMA gift shop, some loud voices in the NFT crowd recast any negative reaction to their work as mere elitist gatekeeping, aligning themselves not only with the irreverent vanguard of the Factory but with leading artists of the 21st century such as Damien Hirst and Maurizio Cattelan. And while Cattelan, too, has drawn the crucial distinction between the work (the deed) and its malleable expression (the banana), in “Strange Rules” the tone is not one of mischievous irreverence—the glee of what one got away with—but of intense care. These people are nerds, and they want their obsessive labors recognized.
The French philosopher Michel Foucault predicted that one day “the author function will disappear […] behind all these questions, we would hear hardly anything but the stirring of an indifference: What difference does it make who is speaking?”
Has that day arrived? I’d argue the important question is a different one.
As works that were authored are pressed into a machine that produces an authorless average, and as dissemination technologies like the internet inundate us with language that may or may not come from a bot concocted by a PR firm, we now find ourselves obsessed not just with who is speaking, or what is speaking, but—more importantly—with this: “Is it worth listening to?” The answer to this question is not found, according to the framework of Protocol Art, in sussing out the nature of the machine but rather the mind manipulating it.
