There is a painting most of us carry in our heads without ever having stood beneath it. Two hands, two index fingers, and a gap so small it feels like a held breath. Between 1508 and 1512, on the ceiling of the Sistine Chapel, Michelangelo placed the whole story of creation in that gap. Not in the bodies, not in the clouds, not in the crowd of figures gathered around God. In the few centimeters where two fingers do not quite touch.
The image at the top of this Note borrows his composition and turns it around. A human hand reaches towards a hand that was made, not born. Between them hangs a single red thread. This time it is not a god giving life to a person. It is a person giving form to a machine.
I did not get that image on the first try. I asked two different tools, again and again, for one hand, one thread, one horizon of light. I received hands with too many joints, threads tied in knots, warm light where I wanted cold, glass that looked like plastic, and one beautiful version that belonged to another story entirely. The machines were generous. They offered me many hands. None of them could tell me which one was right.
That part was mine.
And even the one I chose was not quite what I had seen in my mind. Close, but not mine yet. So I took it into Lightroom and finished it by hand, until the light, the texture and the red of the thread matched the image I had been carrying. What you see at the top of this Note is neither the machine's work nor mine alone. It is both, and that difference is exactly where authorship begins.
And that is the question fashion keeps circling, usually too fast:
Who is the author now?
The machine was born from a thread
We like to talk about artificial intelligence as if it had arrived in fashion from somewhere far away, from laboratories and server farms. It did not. It came home.
In 1804, in Lyon, Joseph Marie Jacquard presented a loom that could weave complex patterns on its own, guided by chains of punched cards. Each card was a row of decisions: lift this thread, leave that one down. A few decades later, Charles Babbage borrowed the idea of punched cards for his Analytical Engine, the machine now remembered as an ancestor of the computer. In 1843, Ada Lovelace, who wrote what is often called the first computer program, described it in a sentence I cannot stop rereading:
“We may say most aptly that the Analytical Engine weaves algebraical patterns just as the Jacquard-loom weaves flowers and leaves.”
The computer learned to think by watching cloth being made.
I grew up close to the other end of that thread. My parents ran a garment factory, with hundreds of people working at the sewing machines and thousands of pieces leaving the floor. I was very young then, but that is where I first saw scale: rows of machines, bundles of cut fabric moving from hand to hand, and the quiet discipline it takes for the thousandth piece to be as good as the first. Years later, with a small label of my own, I learned the other half of the craft: how to carry a vision all the way into a real garment. How to explain an idea to the people who will cut and sew it, how to hold the quality line when a deadline pushes against it, how to stand at the end of production and decide that a piece is right. And I learned that a garment holds the time of the people who made it. The better it is made, the less you see the making, and yet you feel it.
So when the machine now returns to fashion to make not fabric but faces, it is not a stranger. It is a descendant coming back to the house it was born in. And like every descendant, it brings the old family argument with it.
That argument is not new either. Between 1811 and 1816, English textile workers known as Luddites broke into workshops at night and smashed the machines they believed would erase them. Two centuries later, models, photographers and stylists watch synthetic campaigns with the same fear, expressed in comment sections instead of with hammers. The fear was never really about machines. It was about whether anyone would still need the hands.
The face that was never born
In August 2025, readers of American Vogue turned the page to a Guess campaign: a young woman, calm and flawless. In the corner, in small print, one line: “Produced by Seraphinne Vallora on AI.” She had never existed. A TikTok video pointing at that line was watched more than two million times. The anger that followed was not really about one advertisement. It was about the feeling of having been looked at by someone who was not there.
A few months earlier, H&M had done something that looked similar and was, in fact, the opposite. It created digital twins of thirty real models. One of them, Mathilda Gvarliani, described hers as “just like me, but without the jet lag.” The models keep control over how their twins are used and are paid for them the way they are paid for photographs. The images were marked as AI. Nobody disappeared. A person was extended, not replaced.
Then, in February 2026, before its show in Milan, Gucci published AI-generated images labeled “created with AI.” Some saw a house experimenting in public, which is what great houses are supposed to do. Many others saw “AI slop” and asked why a name of that stature would choose a cheaper tool over real people. The disclosure was honest. The question it raised was not about honesty. It was about what a great name is expected to carry.
The same technology, three different stories. One erased a person. One extended a person. One tested how much a name can hold.
Images not made by hands
There is an old word for images like these. In the Byzantine world, the holiest images of all were called acheiropoieta: “not made by hands.” The Veil of Veronica, the Mandylion of Edessa. Faces believed to have appeared without any painter, imprinted directly, untouched by human intention. For centuries, the absence of a human hand was the highest proof that an image was true.
Today, the same absence is the first reason we stop believing.
What changed is not the image. What changed is who we imagine standing behind it. For the faithful, the invisible author of an acheiropoieton was God, and the missing hand was a promise: someone vouches for this. When we look at a synthetic face, the missing hand promises nothing. There is no one behind the glass. No one who was there, no one who agreed, no one who answers.
We have always trusted images the way we trust people: by asking who stands behind them.
Luxury has always sold exactly that. The hand-finished seam, the atelier, the signature, the name on the door. When we pay more, we are rarely paying for the object alone. We are paying for the proof of a person behind it. That is why the backlash against synthetic campaigns is rarely about technology. It is about a promise that quietly changed without anyone saying so.
Authenticity as the new rare material
In 1935, Walter Benjamin wrote that what withers in the age of mechanical reproduction is the aura of a work of art: its presence in one particular time and place. Photography was supposed to kill the original. Instead, it made the original more precious. We now live at the far end of that road. Retouching, filters, beauty modes, perfect skin on every screen. AI is not the beginning of this journey. It is its logical conclusion: a reproduction that no longer needs anything to reproduce.
And something interesting happens at the end of a road. The more synthetic images surround us, the more value moves to what cannot be generated: a real body in real light, a real hand that made the thing, a moment that actually happened, a person who agreed to be seen. Scarcity does not disappear. It changes material.
For luxury, authenticity is becoming what gold once was.
Two continents, one direction
The law is arriving at the same conclusion from both sides of the Atlantic.
In New York, the Fashion Workers Act has required, since June 2025, clear written consent from a model before a digital replica of them is created or used, specifying its scope, its purpose, the pay and the duration. California has introduced its own protections against unauthorized digital replicas of performers. In the European Union, the transparency rules of the AI Act have applied since August 2026, including the duty to disclose deepfakes.
Seen from Europe, the pattern feels familiar. We tend to write the rule before the thing scales; America tends to write it once the market has tested the thing. Europe protects the original because it has spent centuries living among originals: cathedrals, workshops, guilds, houses that outlived their founders. America protects the person because it built an economy on likeness: the star, the face, the endorsement. The routes differ. The destination does not. Consent and disclosure are becoming the floor, not the ceiling. A brand that works on both sides of the ocean will end up working to the stricter standard anyway. It might as well choose it, and say so.
The red thread
In several traditions, an invisible red thread connects people who are meant to meet. It can stretch and tangle, but it does not break.
The thread in the cover image is the same thread that ran through Jacquard's loom, through the machines on my parents' factory floor, through every seam that someone once finished by hand. It does not rise like a beam of light. It hangs. It has weight. Someone has to hold each end.
That is the part no tool can generate. The machine can produce the hand, the light, the glass, the horizon. It cannot decide where the thread goes. It cannot decide what an image should mean, whom it serves, what it must never pretend to be, and which of a hundred beautiful versions is true to the house that will sign it. Those decisions are still human. They are, quite precisely, the work of art direction.
Where AI belongs, and where it doesn't
AI belongs in the sketch and the mood board, where it can test forty versions of light before a single crew is booked. It belongs in worlds that are openly imaginary: dreams, impossible sets, garments that could never exist and do not pretend to. It belongs in extending a real person, with their consent, their control and their pay. And it belongs in the quiet production work that frees budget and attention for what matters most.
It does not belong where it replaces a face while pretending nothing has changed. It does not belong in the promise of craft: if a brand sells the hand, it should not fake the hand. It does not belong in images that present themselves as a record of something that never happened. And it does not belong in the small print in the corner.
In practice
Before an AI image goes live, I would ask three questions.
Who is this person? If the answer is no one, say so clearly, where people will actually see it.
Whose likeness, whose consent? Written, specific, paid and limited in time. The New York standard is a good default anywhere.
Does this image promise something the product cannot keep? If you sell craft, show craft.
The loom did not end weaving. It changed who could weave, and what a hand was for.
The future of fashion imagery will not be decided by what machines can make.
It will be decided by what we are willing to sign.
xx,
Laura
If this note stayed with you, you might also like The Hands We Do Not See, on slow fashion, craft and the people behind every seam.
More on: AI · Fashion · Technology · Creation
