The rapid rise of face recognition technology is undeniable these days. In some parts of the world, it is already omnipresent and used for a variety of purposes such as the identification of passengers at airports, the surveillance of citizens in public places or even criminal convictions. While Artificial Intelligence (AI) had not been able to distinguish a blueberry muffin from a dog (specifically a chihuahua) not that long ago it, too, is advancing.

However, it is still flawed and face recognition results in an average rate of 3% failure. This may not seem like an awful lot, but considering a crowded place, such as the busy streets of New York or an airport, these mere 3% may quickly accumulate a myriad of mistakes each day, potentially, resulting into detrimental consequences. Fundamental human rights are at risk. Automated face recognition technology is a highly controversial topic – from a legal, but also from a technical point of view making the urge to protect one’s data increase rapidly.

A fashionable approach to data protection

As we all know, ChatGPT and other applications bring about a vast variety of new possibilities. AI used to design clothes being just one of them. Rachele Didero, a fashion designer from Italy, did just that. She founded Cap_able – a company that specializes in AI-designed patterns, but those are not just any patterns. They are made to protect the biometric data of its wearer by making it impossible for face recognition technology to function properly. Therefore, keeping the identity of that person a secret. Their garments aim to be ‘The GORE-Tex of Data Protection’, she recently proclaimed at the Imagination In Action’s summit, called ‘Forging the Future of Business with AI’ as can be seen here on YouTube. They want to stand out and raise awareness about the risks for fundamental rights due to facial recognition and contribute this way to the debate on the legitimate use of AI.

So, how does it work?

At first glance, Cap_able’s flamboyant clothes may look like a remnant of the 70ies: super colorful with wild busy patterns and flared bell-bottoms.

Yet, there is a method to the madness. The fashion house tested their garments with the neural net-based object detection system YOLO (You Only Look Once), designed to identify humans and objects alike in real-time like shown here on the company’s YouTube channel. This range of, in fact, cutting-edge patterns uses a variety of special cues and details. They include a so-called ‘adversarial patch’ on a specific type of polychrome fabric, which is able to fool AI-systems into identifying humans as animals. So confused by the fabric, AI does not even scan the human’s face and its features. This also makes protective face covers and glasses to avoid identification obsolete. As a result, biometric data can neither be detected nor extracted and then processed further for any connected data bases.

Outlook

With its bright colors and its price tags starting at 370 euros a pop, this fashion line might not be everyone’s cup of tea. Yet, there is a silver lining: it comes to show that we are in fact still capable of protecting our personal data in this day and age. It might not always look like what we have imagined, but I am sure with an open mind the future holds a lot more exciting new technologies to come in this regard.

| | | , , , , | 3 Kommentare