In 1925, Ernst Friedrich founded the Anti-War Museum in Berlin, which he intended as a collectively curated institution, where everyday objects and candid photographs documenting the atrocities of war would be exhibited. His efforts stemmed from a clearly defined belief, shared by many at the end of the First World War, that looking war in the face would strip it of its allure. Before the advent of photography, he said, we lacked real evidence of war’s horrors. With photography, “the mask has been torn away … He who then still believes in this mass butchery, let him be locked up in a mad-house.”1 Yet, like many other conscientious objectors of his time, it was Friedrich who was locked away in a mental hospital for refusing to join the invading German army in its aggressive war of expansion, and then convicted for sabotaging military activities in 1917.2 In the coming decades, his museum was repeatedly vandalized by the Nazis. He fled persecution by the secret police, saved numerous people from death, and helped countless others. At least until the late 1940s, when he attempted to create a new anti-war museum in Paris, he seemed to have maintained some belief in the power of images to expose and ultimately stifle violence. Can such conviction be sustained today?
A mere century after Friedrich first founded his museum, we have no shortage of corroborations, nor of portrayals of war. The internet is akin to a perpetually expanding exhibition of war, akin to a labyrinth of violence constantly spawning new rooms. Documentary footage of war has thoroughly saturated everyday visual culture. On social media, new evidence of war crimes unfolds in a manner both casual and horrific, among adverts, spam, pornography, memes, and cute pet videos. Online, at a speed that prohibits us from fully understanding what we are seeing, posts about someone’s latest career achievements and personal joys appear on our screens interchangeably with someone else live-streaming their own death. Whole channels, fora, and group chats are devoted exclusively to circulating war materials, with graphic footage accessible openly and daily. We cannot anymore say that we lack records, or that we need better, more truthful representations.
War photography’s original intent to demythologize the battlefield by showing irrefutable evidence and damning truth falls flat today, because little is hidden. How can we then truly see war in a culture of omnipresent seeing?
This is the landscape in which Alexey Yurenev’s projects unfold. A photographer and visual researcher, the cornerstone of his recent body of work titled Silent Hero is a series of AI-generated images. Drawing on a dataset of tens of thousands of public domain images from World War II, ranging from landscape shots to portraits of soldiers, these hallucinatory depictions disfigure what is familiar, just as war itself does. From a distance, and at first glance, his AI-generated black-and-white landscape shots could pass for photographs. Yet, upon a closer look, we notice how figures resembling humans, animals, and machines fuse into a single blur that captures the violence of war, where bodies melt into the soil and technology penetrates flesh. The AI portraits of soldiers similarly dissolve any illusory moral clarity lingering in the “originals”—photographs in which soldiers pose with heroic dignity, which were used as the dataset. In the synthetic version, the portraits do not seem to depict any specific individual. Rather, multiple faces blur into one another, their features folding and collapsing until the human form gives way to something else entirely—distorted, and almost monstrous. These hallucinatory faces appear wounded, cut up, disfigured, as if wrung through the grinder of history, coming out shattered, partial, and punctured through.
They are not a likeness but a prophecy, as if the inevitable trajectory of war has begun to surface, revealing the latent potential for disfigurement and ruin. Artist Hito Steyerl wrote about AI-generated images as “mean,” not only because they tend towards the grotesque, but also because they average things out; in this case, they are a median portrayal of trauma.3
Aesthetics: The Authorless Eye of War
These images were incorporated into Seeing Against Seeing, an artist book Yurenev created in collaboration with designer Teun van der Heijden and the Anti-War Museum in Berlin, which reopened in 1982, fifteen years after Ernst Friedrich’s death. In the book, they are inserted between the pages of Friedrich’s original 1925 manifesto, War Against War.
There, he writes about the persistence of war, “the most diabolical, the meanest and lowest of all crimes of the state … All the treasury of words of all men of all lands suffices not in the present and in the future to paint correctly this butchery of human beings.”4 Where words cannot suffice, photography, in its assumed self-explanatory power and ruthless objectivity, was supposed to do a better job. Yet, the earliest war photographs already distorted reality. They date back to the Crimean War of 1853–56, on the territory of modern-day Ukraine, which is also under occupation today. Back in the nineteenth century, British photographer Roger Fenton was commissioned by his government to produce positive images of the battlefield: noble landscapes, glorifying portraits of soldiers and camps, and poetic, morale-boosting aftermath scenes of cannonballs scattered across roads.
In contrast to such wistful half-truths, the photographs included in Friedrich’s anti-war book are horrific. In 1925, they must have appeared especially disturbing, but even to today’s viewer, who is more routinely exposed to graphic imagery, they remain incredibly impactful. There are shots of scattered corpses, close-ups of men severely disfigured or operated on without anesthesia, and of their torn-apart faces with gaping holes in them.
These photographs disturb us not only because they reveal suffering but also because they almost defy our brain’s ability to categorize what we are looking at as human faces. Though they hail from a century ago, they resemble AI-generated footage of today. But they are not fake. Quite the opposite, they capture horrors so real that our perception scrambles upon the encounter, struggling to comprehend what it is that we are looking at. When confronted with these photographs, our mind’s facial recognition system malfunctions. Between these original images, which already look unreal, and the AI-generated reinterpretations, our psyche must recalibrate its definition of what counts as a truthful portrayal of war.
Photography and AI might seem diametrically opposed. Where one is considered the gold standard of evidence, the other is accused of “hallucinations” and falsifying reality. Yet photography, too, was once dismissed as unnatural, blasphemous, and mechanical.
Described as an offense to human art, photography was considered authorless, to the point that early photographers were ridiculed as “sun worshippers,” as if the only real artist at work in making a photo was sunlight.5 Photography was additionally considered blasphemous and base. Only God should be able to freeze time, proclaimed a critic in 1839, “not any machine of human devising.”6 This brings to mind how artificial intelligence is often criticized today: inhuman, inauthentic, fake, vulgar, and a travesty to human feeling and art. In the harsh words of Hito Steyerl, AI images are “demeaning” and “converge around the average, the median; hallucinated mediocrity”; “in style and substance they are “mean images.”7 To be the mean or the average is not exactly an aspirational quality. Yet, in one of philosophy’s most enduring texts, Aristotle’s Nicomachean Ethics, “the mean” refers instead to the idea of the virtuous middle between excess and deficiency.8 Courage, for example, is the mean between recklessness and cowardice.
AI images might also be this sort of mean: an attempt at distilling a common, and therefore noble and meaningful, form. This is especially relevant in Yurenev’s synthetic war images, made using generative adversarial networks (GANs). In this method, two neural networks—one that generates images and another that judges them—compete with each other. The generator learns to create images that seem real, while the discriminator tries to spot the fakes. Typically, this process helps the generator strike a balance between realism and originality. In Silent Hero, it is the artist’s hand that sets the parameters for what is a desirable image, entering as a third person into the process of averaging out the data. Either way, the generator must avoid outputs that are either too uniform or too chaotic, aiming at a certain mean. In this sense, Silent Hero inherits photography’s unstable relationship to authorship, where both the machine and its operator contribute to creation.
Memory: Synthetic Inheritance
What does it mean to generate images of war, not as documentation but as a sort of synthetic inheritance of things seen and unseen, latent and explicit? As photography becomes mundane and abundant, could these generated visuals still capture an existential truth about war?
Unlike traditional photography, AI-generated images don’t reference specific events or faces, but rather create composites. Though lacking singular referents, these delirious, dreamlike images drawn from the collective database reveal a truth: that when one wants to really perceive war, perception and memory become blurry and unstable.
Perceiving, recalling, seeing, or witnessing war is a cognitive task that can border on psychological trauma, what Asia Bazdyrieva and Svitlana Matviyenko have recently called “the labour of witnessing.”9 As the writer Han Kang confesses, researching in the archives for her novels Human Acts (2014) about the Gwangju uprising in 1980, and We Do Not Part (2021) about the Jeju massacre of 1948, caused her relentless nightmares. This is how the human brain might be adjusting its grasp of the world in real time, through the unconscious churning out of night terrors.10 Despite such high cost, “the [human] urge to document is something basic, almost mechanical, involuntary.”11 The automatic reflex to document reality grasps at the unreality of war, felt in the injunction and resistance we may feel when looking at it—that this simply “cannot be real.” There is widespread social denial of reality: of the scale of violence and torture happening daily in the world, both in combat and peaceful zones. It is in this context that “looking at war hurts; it works as a trauma, heralding the return of the real.”12 The “real” here is not only the violence itself—whose cruelty often verges on absurdity—but the deeper recognition that safety is a fleeting, arbitrary condition, and that evil can coexist with the ordinary joys of daily life. Like abuse, war doesn’t always arrive as total rupture, but rather seeps into the everyday, transforming it from within. War saturates and distorts daily reality while coexisting with it. Life both does and does not go on.
This impasse—the intrusion of violence into ordinary life, and the impossibility of fully grasping it—is inherited from those who live through war by the generations that follow. In another installment of the Silent Hero series, Yurenev turns directly to his grandfather, a war veteran who, like many, remained silent about his experience, refusing to narrate or visualize it. “I received his medals, but I didn’t receive his memories,” Yurenev remarked.13 Deprived of direct testimony, Yurenev searched for other means of accessing the war that cannot be talked about or forgotten. In the short film Silent Hero – No One is Forgotten, a group of the last remaining Red Army veterans of the Second World War, now living in Brooklyn, New York, are shown the set of AI-generated war images. They dress in their ceremonial military uniforms and display their medals as they view the images and speak on camera about the war. While some claim, “I can’t remember anything” as they examine the AI-generated scenes, others describe what they see in detail: “tanks” and “battles that took people’s lives.” One veteran remarks, “It hurts a lot to look at burned bodies of those people,” while another, his voice trembling, describes how a synthetic image shows “someone running somewhere, from tank to tank.” Though informed of the artificial nature of these images, the veterans seem to perceive real people and landscapes in them. The line blurs—both for those watching and perhaps for the veterans themselves—between descriptions of the AI-generated images and actual memories surfacing while viewing them. “This is terrifying,” one says of the AI-generated portraits. “This is a disfigured face, ears all popped out, [wearing] a soldier’s hat.” “A burned face!” another shouts out.
Some veterans cry while describing the people who were killed. While some memories of war are malleable like digital images, the especially traumatic ones remain frozen, like photographs. The pain of war freezes time and captures its victims in a perpetual present: “I will never forget this horrific image [of murder],” one veteran says. Some images cannot be unseen—they play under our closed eyes forever.
In Silent Hero, the totality of trauma and memory is put through the grinder of machine learning to produce images that both refer to actual events and do not. In a crude sense, they produce something like the median trauma of war photography, as the computer attempts to recreate what is most “probable” about an image of war or about a war-torn face. In a more poetic way, they capture the truth of the unconscious in a similar way to literature and art, by resonating in the prelinguistic spaces of collective memory. They remember for us what cannot be seen but equally cannot be forgotten.
Morality: The Feed of War
Apart from the question of aesthetics and memory, AI-generated war photography has ethical and moral stakes. The question of “realistically” portraying war goes beyond the dilemma of which images can be trusted. At the heart of the question is something deeper: Can we accurately portray the immorality of war? This question of moral accuracy, rather than just visual realism, is also an important composite of truth.
Most often, the morality of AI-generated images is discussed in relation to their visual realism. Amid the concern over propaganda on social media, the whole technology of artificial intelligence comes under intense scrutiny.
Questions of falsity and authenticity are inherent to all media, but especially so in the case of war, which has consistently faced accusations of inaccurate depiction. Contemporary developments such as deepfakes—AI-generated materials that appear completely realistic to the human eye or ear—contribute to the already widespread belief in the (alleged) ancient Greek saying that “truth is the first casualty of war.” This belief itself can operate as a weapon of war, imploring people to “explore both sides” or inviting us to consider that “truth lies in the middle.” Morality can be easily lost in the “fog of war,” when such beliefs provide us with an emotional shelter, barring us from taking decisive ethical stances. The seemingly correct position of saying to oneself, “no to war,” or our desire to reject thoughts about war—or other instances of violence and suffering—as too ugly or pointless, can quickly turn into denial. “No to war” can transform into “there is no war,” or, “it is impossible for me to think about war,” further weakening our ability to understand reality.14
The matter of accuracy has been extremely crucial in the history of war photography, and deeply connected to the possibilities of recording media. The Vietnam War (1955–75) is often credited in Anglophone media history as the first televised war, though earlier conflicts—including the Algerian War of Independence (1954–62), the Chinese Civil War (1927–49), and anti-colonial uprisings such as the Mau Mau rebellion in Kenya (1952–60)—already exposed local populations to the immediacy of war through photography, radio broadcasts, and underground or state media. Today, in the face of AI-generated footage, media institutions and users alike scramble to develop authentication tools, watermarks, and methodologies to verify what is real. Among the contemporary live stream of war footage, artist Mykola Ridnyi talks about a pervasive “blindness—although we live in a world with the best possible access to information, the problem is that there can be too much information, which causes it to manipulate our opinion.”15 Amid this climate of visual doubt, it is no longer clear whether the demand for accuracy can restore a stable epistemology of vision—or whether it simply reanimates an old fantasy: that the camera, or now the algorithm, could ever tell us the whole truth.
Telling the truth is not only a matter of producing objective recordings. In fact, the very first war photography, by Sir Roger Fenton, consisted of objectively recorded landscapes of battlefields. Does this really get at the truth of war? In order to glimpse war, images must confront suffering and violence in a manner appropriate for their times. Today, war crimes aren’t just documented publicly online—they’re often met with gloating, cruel comments right beneath, celebrating the atrocities. On the internet, we can all watch how people openly revel in suffering and violence. Whether on state channels and propagandistic bot accounts, or through the comments and profiles of civilians, we can witness just how much pleasure people take in the violence inflicted upon others, or how openly some celebrate the suffering inflicted in their name. Nothing is hidden away from the eyes of the camera: it shows plainly that “the fascist is not doing ‘something else,’ but doing what he wants to do.”16 When violence and vulgarity become the dominant modes of representation—surrounding us on all channels—ethics might have to retreat into other realms, where the possibility of seeing the horror of war still exists. As images of war and violence become commonplace, shock and shame shift into synthetic imagery, where witnessing takes on new forms. In this task, AI art can be as accurate as photographic realism, becoming a refuge for this sort of truth.
This evolution in witnessing and visual forms is not incidental. War is also where the entire history of technology is visible. It differs from other forms of suffering through its scale and the might of the technologies it employs—beside which human life appears fragile and helpless. It is where human perception undergoes continuous transformation, forced to adapt to the very instruments that assault it.
In the words of Paul Virilio, “the history of battles is above all a history of radically changing fields of perception.”17 There is, for example, a story from the front line about a soldier who operated a thermographic camera and night-vision device, and who after some time started dreaming in infrared.18 His experience reveals something important about human vision, the mind’s eye, and technological perception: our machines, especially the ones we kill with or defend ourselves with, infect and transform how we see. As filmmaker Oleksiy Radynski notices, in response to the World Wars (1914–48), multiple visual avant-gardes emerged in Eastern Europe to capture these changes in perception.19 This witnessing is now continuing with artificial intelligence, a form that befits our contemporary culture of composite, dreamlike, surreal violence that overwhelms the internet today.
Yurenev’s AI-generated images capture reality differently than photography does, but they do capture it. Photography shows us what happened; art shows us what happens. What we see in these AI images is not the appearance of war, but its refusal to appear, even when it is right in front of us. Silent Hero’s dreamlike shots capture the certainty of suffering, while simultaneously unmasking the impossibility of faithfully capturing war itself.
Truth emerges through art as revelation rather than representation—it manifests in the collective inheritance of all that has happened, providing an irrefutable and damning testimony of war’s “average” image.
Ernst Friedrich, Krieg dem Kriege (Anti-Kriegs Museum, Berlin, 1924), 23.
Terrence Malick’s film A Hidden Life (2019) explores a similar case in the context of the Second World War. Franz Jägerstätter, an Austrian farmer and devout Catholic, refused to fight for the Nazis. He was declared mentally ill, jailed, and then sentenced to death.
Hito Steyerl, “Mean Images,” New Left Review, no. 140 (2023).
Friedrich, Krieg dem Kriege, 22.
Lois Rosson, “What is AI Doing to Art?” Noema, April 11, 2023 →.
Cited in Blaise Agüera y Arcas, “Art in the Age of Machine Intelligence,” in Atlas of Anomalous AI, ed. Ben Vickers and K Allado-McDowell (Ignota Books, 2020), 113.
Steyerl, “Mean Images.”
Aristotle, Nicomachean Ethics, ed. Roger Crisp (Cambridge University Press, 2014).
“The Labour of Witnessing,” posted November 6, 2022, by Theater of Hopes and Expectations, YouTube →.
Judith Shulevitz, “Where Han Kang’s Nightmares Come From,” The Atlantic, January 14, 2025; Victoria Kim, “A Nobel Laureate Who Mines Her Country’s Nightmares, and Her Own,” New York Times, January 21, 2025.
Tomasz Szerszeń, Być Gościem w Katastrofie (Wydawnictwo Czarne and Dwutygodnik, 2024), 137. All quotes from this remarkable book, To be a Guest in the Catastrophe, by a Polish visual anthropologist and photographer about watching the war in Ukraine (2014–ongoing), are in my translation.
Szerszeń, Być Gościem w Katastrofie, 143.
“New Histories: Alexey Yurenev,” posted March 21, 2024, by Raclin Murphy Museum of Art, YouTube →.
This argument is paraphrased from Szerszeń, Być Gościem w Katastrofie.
Anna-Lena Werner, “Interview: Mykola Ridnyi,” Art Fridge, November 8, 2022.
Barbara Ehrenreich, foreword to Klaus Theweleit, Male Fantasies: Women, Floods, Bodies, History (University of Minnesota Press, 1987), xi–xii.
Paul Virilio, War and Cinema: The Logistics of Perception (Verso, 2009), 10.
Nicola Abé, “Dreams in Infrared: The Woes of an American Drone Operator,” Der Spiegel, December 14, 2012.
Sandra Biberstein, “To Watch the War: An Interview with Olexii Kuchanskyi and Oleksiy Radynski,” Crisis and Communitas, October 24, 2022.
