Gaby
The Turin-based startup Capable has launched the Manifesto Collection, a line of clothing designed to confuse facial-recognition technology. The vibrantly colored knitwear is designed to fool artificial intelligence systems that identify people.
A person dressed in these bright garments appears to recognition systems as a zebra, dog, or giraffe.
Protection Stronger Than Makeup or Masks
The founders of Capable, Rachel Didero and Federica Busani, developed the collection after nine months of research. During that time, they tested numerous patterns, materials, and knitting machines, as reported by Dezeen. Their innovation was evaluated using the YOLO object-detection neural network, one of the fastest real-time systems.
Ultimately, the Manifesto Collection was born. It features high-quality Egyptian cotton and includes sweaters, t-shirts, pants, and dresses. These extravagant yet practical items don’t require ironing.
The creators say the clothing protects biometric data more effectively than makeup or masks. Wearing it eliminates the need to cover your face.
A hand-knitted sweater from your grandmother doesn’t offer these protective qualities. The face-concealing algorithm is embedded in the fabric’s texture: the bright patterns are designed to make recognition systems see the wearer as an animal.
The research findings formed the basis of Rachel Didero’s dissertation. She hopes the new clothing line will spark a public discussion about the ethics of facial-recognition cameras.
The Manifesto Collection aims to raise awareness about biometric privacy rights among modern audiences. Didero says that the improper use of biometric recognition systems affects much of the world’s population.
The Capable team isn’t the first to try to protect identities from recognition technology. In 2019, Polish designer Ewa Nowak created a brass mask that makes faces unrecognizable to surveillance cameras. In 2020, Brooklyn-based designer Sarah Sallam introduced jewelry and accessories intended to help protect against digital tracking.
