With wearable tech rapidly taking off and Google planning to publicly launch Glass by the end of this year, how do we safeguard privacy when the devices could be recording at any time? A team of researchers from Indiana University have a simple answer: Just teach your devices to recognize when it’s not okay to have the camera running.
The researchers call their solution PlaceAvoider, a currently in-development software application designed to allow users of wearables with recording capabilities to identify places where they don’t wish to record--or worse, upload to the web--by blacklisting them. The team’s co-leader, Apu Kapadia, head of Indiana University’s Privacy Lab, told us about the project’s origins via email:
We recognized the many great applications of wearable cameras, as well as the privacy harms stemming from their use. Given our team's strengths in privacy-enhancing technologies and computer vision, it seemed like the perfect, relevant problem for us to address. In general, trying to have an algorithm discern whether an image is "sensitive" is a hard, if not impossible, problem. So we explored specific types of sensitive images that could be recognized. We realized that blacklisting sensitive spaces would be useful to people and may be a tractable problem to solve. And thus the PlaceAvoider project was born.
Such an approach suggests that the mounting ubiquity of wearable devices dictates a future in which recording is the default state--that, instead of actively having to turn our devices on, we’ll have to consciously remember to turn them off. As such, it’s going to be more difficult to tell when someone is recording you. Google Glass, after all, doesn’t actually have a hardwired recording light.
While software like PlaceAvoider would be a boon to the personal privacy of those who use wearable tech, it could potentially be a reassurance to those around the user--provided there were some way of bystanders knowing that they were "off the record," so to speak.
When asked about this, Kapadia responded that while the issue is an important one, the approach taken by PlaceAvoider in its current state may not be best suited toward addressing it:
We think studying and addressing "bystander privacy" is of great importance, and we are currently exploring this issue as well. Already PlaceAvoider can blacklist spaces from other users ("please don't take pictures of my office"), but this approach requires cooperation from the camera owner to honor others' privacy demands.
However, privacy concerns aren’t the only potential applications for PlaceAvoider. In a story for the MIT Technology Review, co-lead David Crandall states that its algorithms could be used to better automate the organization of vast quantities of images, thereby increasing the utility of image libraries.
Also, having the software in place could be a big help when law enforcement officials accuse you of using your wearable tech to conduct criminal activity.
[Image: Flickr user Prayitno]