There’s something strange about images that should exist but don’t. Like a photographic gap in time, a hole in the collective visual memory. I’ve recently found myself drawn to one particular kind of such image: the precise moment someone damages a work of art. Not the aftermath—the slashed canvas, the splashed paint, the outraged headlines—but that one split second when the act happens. The swing of a knife, the arc of a rock, the sound of a spray can. Oddly, almost none of these moments have been visually documented.
We’re so used to images being everywhere, and though cameras have been around since before the beginning of the 20th century, it's striking that at these moments no camera was around. Take Mary Richardson’s 1914 attack on Velázquez's Rokeby Venus in the National Gallery in London. She walked in with a meat cleaver, slashed the nude several times, and stood by with a sense of political conviction. We have images of the painting afterwards, repaired and in black-and-white newspapers. But not from the actual moment.
Or Kazimir Malevich’s White on White, which was "enriched" by the Russian performance artist Alexander Brener in 1997 with a neon-green dollar sign. There were people in the museum. But again: no photo of the actual act. Just the result. Just Brener later, fists raised, flanked by security, proudly marching away from the scene.
I recently started to write prompts to generate images of these moments with AI. Not as a form of glorification, but out of curiosity: What could these moments have looked like? If you could freeze the precise second someone takes a knife to Barnett Newman’s massive field of red, or throws a rock at the Mona Lisa, what would you see?
To my surprise, the generated images–with support of ChatGPT–weren’t “shocking” in the way I expected. What surprised me instead was how cinematic, how theatrical they felt. The expressions. The body language. The stunned bystanders just starting to process what they’re witnessing; it's probably partly due to how I described the prompt, but the AI probably also "knows" context about the situations I described. Eventually these aren’t just recreations; they’re visual dramatizations. Moments suspended in uncanny stillness, full of tension.
Without actual photos, we’re left with either eyewitness descriptions or post-facto images of slashed canvases behind glass. My attempt to fill that gap—albeit digitally, theatrically, speculatively—opened a new kind of window: not into what really happened, but what we might see if we had been standing there, our phone miraculously ready, before history settled into memory.
I hope to see a future filled with AI-generated digital tableaux vivants—reimagined scenes of breaking news or historically significant moments—that help visualize undocumented stories. Just as paintings once captured decisive episodes long before the camera existed, these modern dramatizations could offer a new kind of visual storytelling: evocative, speculative, and theatrical, yet grounded in real events.