As Al Pacino observes in Scarface: “The eyes, chico, they never lie.” Little wonder, then, that data-hungry Silicon Valley wants to train sensors on our peepers. Eye-tracking cameras, small enough to sit inside a pair of goggles, can detect minute changes in pupil dilation, direction of gaze and focus. If a new generation of virtual reality headsets and smart glasses takes off, millions of us could soon be wearing them.
While it is still early days for this technology, it will undoubtedly prompt questions about privacy. The situation resembles where we were with facial recognition a few years ago. That technology’s potential for undermining privacy was obvious long before it became widespread. Yet it is only recently that tech companies have faced real pressure to address these concerns.
The implications of eye tracking make the objections to facial recognition look almost trivial. Not only can it figure out exactly what you’re looking at, it can, via pupil dilation, assess your emotional reaction. Beyond the field day that marketers will have with this data, the American Civil Liberties Union has warned that eye tracking could be used to detect drug and alcohol use, mental illness and sexual orientation.
There is an opportunity, though, to face up to the threat to privacy now, before the technology takes off. That is the argument being made by one veteran in the field of augmented reality. Avi Bar-Zeev helped develop Microsoft’s pioneering HoloLens headset and, until earlier this year, worked at Apple on an unspecified “new effort” — likely to be the iPhone maker’s smart glasses, which could be released as soon as next year.
Eye data, Bar-Zeev has warned, is so sensitive that it should be protected in the same way that the EU protects biometric data or the US safeguards medical records. “Eye-tracking represents an unconscious ‘Like’ button for everything,” he wrote recently on Motherboard, Vice’s tech site. “The gold rush to your eyes is just beginning.”
There are, of course, legitimate reasons for building eye tracking into headsets. It can be used as an intuitive input mechanism, allowing you to select a digital object simply by looking at it. Or imagine a flight simulator able to identify whether a trainee pilot is paying attention to the right things in the cockpit.
Most crucially, eye tracking will help to miniaturise headsets by making them more efficient. In particular, being able to see where a user is looking inside a virtual world means that processing power can be concentrated in that area, while their peripheral vision need only be rendered in lower resolution, a process known as “foveated rendering”.
Eye tracking is not new but foveated rendering and other applications make it much more integral to VR and AR headsets. Google and Facebook (which owns VR pioneer Oculus) have both bought eye-tracking start-ups, although they are yet to release headsets employing the technology.
But the first devices that do use it are starting to hit the market, including Magic Leap, Microsoft’s HoloLens 2 and HTC’s Vive Pro Eye. These cost upwards of $1,500, so it is unlikely they will be commonplace any time soon. Still, it means app developers have the opportunity to play around with this new toolset.
Relying on consumers to review privacy policies hasn’t turned out to be a great way of protecting our personal data; it seems even less likely to be useful if headset users are unaware that unblinking cameras are surveilling them from inside their glasses.
If AR and VR headsets really are the next big platform, this is one privacy problem Silicon Valley needs to get ahead of.
Tim Bradshaw is the FT’s global technology correspondent
Get alerts on Life & Arts when a new story is published