In Japan, it remains a distinct sign of consideration today. In the West, it became the symbol of an era defined by hysteria, self-preservation, and conformity. Why do we instinctively reject masks in photographs today—and why does Artificial Intelligence nonetheless stick them to our faces as historical anachronisms?

For a Westerner encountering these initially ghostly-looking masks for the first time in the packed belly of an LH 710 flight, the sight induced a strange unease. It took a while to recognize the courtesy and consideration behind them—unlike today, in Japan, these masks were worn to protect others, not oneself.

Then the pandemic hit, and with it came the mass importation of this piece of fabric into the West. But we didn’t import the Japanese humility. We turned the blue rectangle in the middle of our faces into a fetish, a weapon, and a business model.

In photographic terms—to quote the philosopher Emmanuel Levinas—the face is the place where the humanity of the Other appears. It is the primary interface of empathy. The pandemic severed this interface.

To some, the mask became the sacred shroud of solidarity, the visible proof of their own moral superiority (“I follow the rules!”). To others, it became a muzzle, a symbol of absolute state subjugation and a policy that degraded citizens into potential threats.

The fact that behind the scenes, politicians and shady entrepreneurs raked in millions with dubious mask deals, while the actual virological benefit in everyday life often disappeared into the fog of questionable studies, gives the fabric an even dirtier texture in hindsight. Anyone looking at a picture from 2021 today doesn’t just see protection from a virus. They see the profiteers.

In photography, the mask has a fatal characteristic: it is an unforgiving timestamp. A street photograph from 2021, showing people rushing through the city with FFP2 beaks, is not timeless. It is forever trapped in the panic of those years. The Spanish Flu left us with a few photographs featuring masks; Corona literally gave us billions.

Today, we discard these images. We no longer want to see this visual scar. It reminds us of a time of isolation, fear, and ugly faces. The masked portrait no longer has a punctum; it is merely a depressing study of a crisis.

Yet, as we try to erase these images from our collective memory and our portfolios, something absurd is happening in the synthetic world of Artificial Intelligence.

The training data for the large image models has necessarily been fed with precisely these billions of images from the pandemic era over the last few years. The machines have learned: humans often wear something blue or white on their faces.

The machine does not—yet—understand context. It doesn’t grasp that the mask was a temporary phenomenon of a global health crisis. For the AI, the mask has become a statistical feature of being human. This leads to grotesque hallucinations: if we generate a scene today—say, a 19th-century highwayman or a cyberpunk scene—the AI tends to put medical masks on the characters.

The bandit no longer wears the classic bandana to hide his identity, but a clinically clean surgical mask. The AI projects our recent trauma-past into history and the future. It is the ultimate proof of the technology’s context blindness—and perhaps also a little of our own, sometimes blind obedience to authority.

The mask is more than a piece of cloth. It is a relic of a time when the face lost its sovereignty. In analog photography, we try to avoid it as a “flaw in the picture” to restore the subject’s dignity. In AI photography, we have to actively prompt against it, because the machine has internalized it as the default.