Slip on a pair of new invisibility glasses at a party, and in theory, Facebook won’t be able to recognize you the next day when photos show up online. The glasses won’t make you invisible in real life, but they’re designed to thwart facial recognition software.
“We were kind of toying with how people might protect themselves potentially from being recognized when they don’t want to be recognized,” says Tony Anscombe, “security evangelist” for AVG Technologies, which developed a prototype version of the glasses at the company’s innovation lab in Amsterdam.
“Maybe it’s in the background of someone else’s picture at a bar, or maybe it’s a Streetview car going by you and taking pictures that might be included in maps,” he says. “We started looking at how people could protect themselves, and what would be required from a technology standpoint.”
The prototype glasses use infrared lights around the eyes and nose to break the algorithms used to recognize a face. The only catch: Some smartphone cameras include infrared filters that could remove the lights, and cameras that don’t could easily evolve.
“I think it will be an ever-changing world as new sensors and new camera technology comes out,” says Anscombe. “It’s an issue of keeping the technology up to date so it matches the camera.” This version of the glasses also has a retro-reflective coating, so it can bounce back bright light if a flash is used.
While the prototype won’t be on sale anytime soon, the company thinks there’s a clear need for something like this, especially as facial recognition technology improves. If Facebook’s Deep Face tech looks at two pictures, it can tell if they’re the same person with 97% accuracy–basically as well as a real person. The FBI has a new database of millions of photos. And facial recognition may also show up soon in places like retail stores.
“Imagine a scenario where you walk through a mall one day,” Anscombe says. “We already know that some malls are tracking where you walk through your cellular device. Now imagine that someone’s aggregating facial recognition data, so I can identify that it’s you as you walk, and see what you’ve been liking or not liking on a social network. Suddenly I start making you offers. It’s that type of potential intrusion that we believe consumers should have a choice about.”
While the FBI is trying to use facial recognition, so are criminals. Anscombe points to an example of Australian criminals who visited police graduation ceremonies, took photos, and then used facial recognition software to create their own database of undercover cops. “That’s a good use case for this technology, isn’t it–police graduates,” he says.