From a photograph of a room to the phone in your hand, technology is changing what it means to be private.
Consider a photograph you’ve recently taken, something innocuous. Maybe it’s a photo of your meal, or a picture of an object in your living room. Have you ever stopped to think about the background of that image and what it may reveal about who you are, or even where you are?
That doesn’t necessarily mean personally identifying information like discernible documents in the background, or landmarks that might exist outside your window, or even location metadata embedded into the photograph. Being able to identify you from the images you take has long been an elaborate story of what we, knowingly or not, release out into the world, but things are getting a lot more complicated in the realm of personal privacy these days.
What if that innocent photo you took and posted to your various social media platforms were used to find your exact address just by looking at interior features of your home?
That’s exactly what developers at Graylark Technologies have been doing with their newest product: Raven (formerly GeoSpy).
Essentially, Raven is able to use machine learning to look at a photograph of the inside of your house, identify distinguishing features of the architecture and other elements of the home, and match visual features against an existing collection of images associated with known locations. And the kicker is that it’s comparing your photo to images not necessarily even taken by you. They could potentially have been taken and uploaded by previous occupants or even real estate workers.
An example given by Graylark is this photo of a laundry machine, which was used to identify the address of the home in question without using metadata or other information.
Both the Los Angeles Police Department and the Miami-Dade Sheriff’s Office have purchased access to this technology, ostensibly as part of their anti-human trafficking operations.
The technology is still in its infancy, and both researchers and police have reservations about its ultimate usefulness, as it depends on the availability of correlated images, but its development raises some rather uncomfortable questions about the right to privacy.
But if a photograph of a washing machine, something that contains no obvious identifying information, can potentially reveal a person’s address, what can be learned from technologies that collect information continuously, often without requiring us to take any deliberate action?
The Wi-Fi you use and depend on every day can be used to gather surprisingly specific information about both the layout of your home and your position inside it without anyone actually looking inside.
This isn’t to say that your router is actively spying on you all the time. It isn’t. It’s just that with the appropriate equipment and conditions, researchers have demonstrated that wireless signals can be used to infer the layout of indoor environments and even track people’s movements and body positions through walls, without relying on conventional cameras.
Even the phone in your hand is being used to shrink your real privacy and anonymity. And not necessarily in the way you imagine. We all know that our phones have huge potential to betray our personal information to outside actors, and that’s why privacy permissions take up so much space in the setup process.
Every app we download must agree to limit what information it collects and how it shares that information, and those apps can potentially collect an enormous amount of information about who we are, who we associate with, and where we exist. Indeed, tech companies like Meta frequently use that information in advertising and algorithmic recommendations.
But there is also a lot of incidental data regularly collected by our phones that can potentially wipe away any expectations we have for privacy. Even if you have location services turned off, your phone’s use of cell towers to communicate provides the means to locate it in the real world; there have long been legal exceptions to the reasonable expectation of privacy in order to use that data for investigative purposes in certain situations.
These three examples expose a critical difference between privacy and anonymity. Privacy has traditionally been limited by the information we expose to the outside world, whether through deliberate choices or through policies made by the services we use that guard against the collection of information without our knowledge and consent. Anonymity, on the other hand, has been affected by the difficulty that exists in correlating publicly available information with personal identity. Historically, we’ve enjoyed a degree of anonymity simply because identifying someone from scattered information required considerable effort. This technology is dramatically reducing that effort.
Consider your photograph above.
You can publish a photograph anonymously, without including your name or any other identifying information. You’ve deliberately made that photograph public, so the image itself is no longer private. But if Raven can associate that photograph with a particular address, it can compromise your anonymity.
Similarly, wireless sensing can potentially intrude upon your privacy without ever identifying you by name.
And at one time, you could walk down the street, browse in a shop, and go about your day without that activity being directly linked to your identity. Now the technology we carry on our person, facial recognition cameras and plate readers, and the connected point-of-sale apparatus in the store all work to connect your behaviour to an identity profile. A profile that can be used to track, analyze, and measure who you are.
While the associated databases aren’t necessarily connected at present, these technologies are blurring the lines between the concepts of privacy and anonymity. By using information we haven’t conventionally thought of as identifying and mixing it with other, seemingly unrelated streams of information, they provide a result that previously would have taken enormous effort to gather and correlate to the individual. Technology is making it much easier to bring together disparate sources and glean new information from the aggregate.
But what happens when AI agents are given access to user data?
One is reminded of the 2008 action thriller Eagle Eye, starring Shia LaBeouf and Michelle Monaghan, wherein a malevolent super-intelligent AI system gains access to an enormous field of data through cell phones and computers and uses that information to blackmail several individuals into carrying out its real-world plans.
Of course, that was just a movie. The real world doesn’t necessarily work the way Hollywood would have us believe. But it doesn’t take a computer science engineer to see the parallels between that fantastic story and the potential our current technology has to violate the privacy and anonymity of real people.
With systems that excel at gathering and connecting disparate data sets, as modern AI is becoming better and better at, can we still expect the same degree of privacy and anonymity? This isn’t to say that AI, as we currently know it, has any plans to violate your privacy or to use your information in some nefarious scheme, but as mentioned previously, there is always a human presence behind everything an AI agent does.
And all of this begs the question of how exactly we’re supposed to protect ourselves in the face of this growing landscape of technology that can read us better than we can.
It’s tempting to make a comparison to the surveillance state in George Orwell’s 1984, and indeed, there are some striking parallels. But where Orwell envisioned surveillance technology forced on the public by the state as both a means of watching and intimidating the individual, today, we’ve largely adopted these technologies of our own accord. We voluntarily give our data away in exchange for convenience, communication, entertainment, and commercial benefit.
We spend billions of dollars on cell phones annually, gleefully accepting the potential for our data to be shared and used in ways we can’t anticipate.
Wi-Fi is nearly ubiquitous in modern life, and we scarcely even know it’s there until it goes down.
We’re surveilled constantly by a bouquet of disparate technologies whose data pools can, and sometimes are, combined to make a better identity profile. Although some are wont to characterize our society as a police state, we aren’t yet at the point of totalitarian surveillance in the greater world.
But that isn’t to say the day won’t come. We’ve seen shifts in the political landscape toward that ideal, particularly in the United States, and it seems that there’s little standing in the way of bad actors using our data to enact a more oppressive environment. Even in Canada, the government’s lawful-access proposal, Bill C22, which would expand law enforcement access to communications and associated data, stands to pull back some long-held beliefs about what privacy guarantees the government offers.
The checks and balances needed to ensure our reasonable right to privacy are being eroded from both directions. Not only is the technology beginning to blur the lines of what can and can’t be used to identify who we are and what we do, but the governmental systems in place to protect us are apparently not as impenetrable as we once believed.
One might be forgiven for thinking that all of these databases can be connected, providing an almost endless pool of information to be used by powerful people to do whatever they might, and we may eventually see just that. But the situation today is still manageable, and there is still space for regulation to step in to protect what should reasonably be private. Though the argument about what specific information should be protected and what should not is ongoing.
In any event, when an innocent photograph of your washing machine can be used to determine your address, doesn’t that suggest that we’re already building the infrastructure for a surveillance state? Is political will the only thing standing in the way?
Leave a Reply