The Internet of Things is bringing with it an existential shift in how we think about connectivity and the way in which the physical world fits into the digital one. Nowhere is this more readily apparent than the vast network of internet-connected security cameras being paid for and assembled by homeowners and businesses all across the country. As an increasing number of these cameras are being plugged into centralized cloud monitoring services, they raise serious privacy concerns, including the possibility of national-scale facial recognition and the enrichment of digital behavioral and interest datasets with our offline behaviors. As we rush towards this brave new Orwellian future, it seems 1984 is ever closer.
Home surveillance camera networks were once primarily the realm of the wealthy or paranoid and were limited by their inability to do more than archive activities onto local tape recordings that could only be consulted long after an event of interest had taken place. They were primarily for obtaining evidence after a crime had been committed rather than preventing it in the first place.
In contrast, today’s cloud-connected cameras are becoming steadily more ubiquitous as homeowners are lured by their cheap prices and ability to monitor their homes from anywhere in the world.
Most importantly, today’s intelligent cameras can do more than record a crime for review after it happened. They can alert homeowners to suspicious behavior before anything actually happens. Most of today’s cameras come equipped with motion alerts, allowing homeowners to receive a phone alert when motion is detected on their front steps, their side garage door or their backyard and watch video of the intruder live. Burglars attempting to make their way in a broken side window will find themselves not only triggering a live alarm to the homeowner, but having their entire escapade recorded as evidence, all with sufficient time for the police to halt the crime in progress.
Yet as exterior cloud-connected cameras become more popular, they also raise unique privacy concerns society has not confronted before.
Street-facing cameras, including the growing popularity of doorbellcameras, make it possible to remotely watch the public space in front of a house, including both the sidewalk and street in suburban areas not traditionally covered by the camera networks more common in the urban interior.
Given the centralized cloud services to which these cameras are typically connected, it would not take much work for manufacturers to offer real-time facial recognition of known criminals and alerts of suspicious behavior, such as an individual repeatedly casing a house over a period of days or appearing to follow its residents.
In turn, access to such camera networks could be sold to law enforcement agencies under lucrative contracts that would perform real-time facial and license plate recognition of outstanding warrants, known criminals, weapons, suspicious behaviors and other activities of interest to police.
Indeed, at least some companies have already filed for patents, floated product ideas or otherwise made known intentions to do precisely this.
What happens when the public space is saturated by privately-owned cameras performing everything from facial recognition to behavioral analysis?
The demographic bias of current technologies means such systems are likely to exhibit substantial discriminatory behaviors. It is also unclear how municipalities might address the civil rights issues of such systems.
What rights does a user have who finds themselves repeatedly stopped by police due to a particular manufacturer’s facial recognition product incorrectly identifying them as a wanted violent felon every time they walk down the street? It is unclear if such an unfortunate doppelganger even has the legal right in most jurisdictions to actually force the manufacturer to correct their product to prevent those false matches.
Of course, in today’s data-drenched world, it is unlikely camera manufacturers will be content to just sell cameras when the data collected by those cameras is so valuable. It is almost a forgone conclusion that manufacturers will eventually build into their terms of service the right for them to run their own facial recognition, behavioral profiling and other analytics in order to generate rich offline profiles of their owners to resell to advertisers and data brokers. After all, a doorbell camera manufacturer that knows precisely when you come and go from your house and the identity of every visitor you’ve ever had and the time, duration and history of their visits, along with every person who has walked or driven past your home, has an incredibly valuable dataset to sell.
One could even imagine companies giving away their cameras for free in return for the right to collect their data, much as social media platforms do today in the digital realm.
Putting this all together, as the physical world becomes enveloped by the digital, the same surveillance state that watches us online will come to surveil us offline.
In the end, 1984 is ever closer. The only difference is that this time we’re building it ourselves.