Oyster Beauty: Exploring AI Disguises In ‘Dazzle’ By Georges Antoni For Oyster #117

This article originally appeared in Oyster Magazine Issue #117: The Unreal Issue, which you can order online at the Oyster Magazine Print Shop.

***

We’ve all seen them: elf ears, heart eyes, filters that envision how we’ll look in 2069. But what happens when you can apply IRL beauty techniques that look like social media filters but instead act as a kind of digital disguise?

“Your face is being uploaded to a database that could, one day, be sold to the highest paying customer.”

As companies continue adopting facial recognition software under the guise of security, what they’re really doing is invading our space — even when we’re offline. These “AI-based safety solutions,” as they’re often referred, function in a sort of grey area where the law hasn’t yet caught up with technology and our desire for continual advancement.

Sure, while no one else might be able to open up your locked phone screen, your face is being uploaded to a database that could, one day, be sold to the highest paying customer.

Take, for instance, the FaceApp Challenge that took over the internet earlier this year. Within a few days, just about everyone from Miley Cyrus, the Biebers and your dad, had all downloaded the application and posted selfies enhanced — or diminished, depending on how you look at it — by an old age filter that made everyone look over 85.

And just as quickly as the application went viral, social media and mainstream news sites were flooded with reports that the filter was actually giving Russians access to your face for some nefarious project. Since then, the information has been ruled a conspiracy — or maybe not. After all, we don’t know where our photos and faces go once our phones have been turned off (or the next face-changing filter goes viral).

That’s precisely why people have started to push back. But what can you do if you don’t want to bring a hammer to your phone? If you’re not totally adept when it comes to beauty, you can take a page from the face-changing filters themselves, creating hair and makeup just extravagant enough that it provides a kind of technological safeguard when it comes to facial recognition software.

The idea comes from artist Adam Harvey and his project CV Dazzle. Named after a type of World War I naval camouflage, which used Cubist-inspired designs to break down the visual continuity of a battleship and thereby conceal its orientation, the project uses avant-garde makeup, hair extensions, accessories and gems to manipulate the data read by AI algorithms in particular types of facial recognition software.

It takes advantage of their fallibility — the bold patterns distort the expected features targeted by the programs, creating a high tech camouflage that’s also high fashion.

“Self-learning AI relies on high quantities of high-quality data to improve accuracy, so high volumes of unreal data can flip it the other way, perhaps making it too unreliable to even take seriously.”

It might also be the answer to stopping facial recognition in its tracks: self-learning AI relies on high quantities of high-quality data to improve accuracy, so high volumes of unreal data can flip it the other way, perhaps making it too unreliable to even take seriously.

Practically speaking, the goal of beauty — and a curated social media presence, in general — is to create our own self-styled masks. So if they can protect us from potentially dangerous AI, that’s just a bonus. And if we’re already living in a sci-fi horror flick, why not dress the part?