Roth is a senior financial analyst and attorney. He lives in Portland.
What’s the scariest thing, nowadays, about going out in public? One very reasonable answer sounds like an old schoolyard taunt: “your face!”
That’s because going out increasingly subjects you to facial recognition systems. These systems biometrically identify you by matching your unique facial dimensions against huge databases of photos.
What’s wrong with that? A threshold problem is that they’re imperfect—at least for now. For example, many have elevated error rates when identifying dark-skinned people. Also, the technology is often deployed improperly, which may cause it to finger innocent lookalikes. Such problems plague both law enforcement and private-sector applications. For example, what’s your recourse if you’re bounced from a store because you’re deemed suspicious by a facial recognition gatekeeper, like the one reportedly installed recently at a Jacksons Food Stores location on Southeast Grand Avenue?
But perhaps the worst problems with facial recognition are not its bugs (in software parlance), but its features. At its core, facial recognition enables a society of continuous mass surveillance. In China, it serves a repressive, nation-sized panopticon that crushes dissent and persecutes the country’s minority Uighur Muslim population, as multiple media outlets have reported.
Systems in the United States are far behind, but rapidly proliferating. As Georgetown University’s Center on Privacy and Technology has documented, Detroit has developed a network of more than 500 cameras across the city in public locations, including at churches, clinics, and social service organizations. While they protect against street crime, they can also feed images in real time into the police’s facial recognition system. That system is also envisioned to be capable of receiving video feeds from other sources, such as drones and police body cameras.
Detroit has obscured the program from public scrutiny, so it’s unclear exactly how and to what extent law enforcement is currently using its facial recognition powers. What is clear is that its capability is large and growing. It’s also clear that any system that has the capacity to effectively create 1984 2.0 demands public oversight, accountability, and a robust debate about its appropriate reach. Chicago has a system that may be even more extensive than Detroit’s, with some 20,000 cameras. Like Detroit, Chicago has successfully obfuscated the extent of the system’s use.
In Oregon, the Washington County Sheriff’s office began using facial recognition in 2017 (“Amazon’s facial recognition technology is supercharging Washington County police,” April 30). The county is quick to assert –presumably with a straight face – that it uses the technology only in a benign, non-Big-Brotherish manner. However, in the absence of rules or oversight, we’ll mostly have to take their word for it. Meanwhile, Rekognition,the Amazon-owned system Washington County uses, recently added “fear” to the list of (now eight) emotions it can detect on the faces of surveillance subjects. Who decided this creepy innovation was necessary? We also have very limited information about what other Northwest government or private entities are using facial recognition, because there’s nothing to stop somebody from scanning your face behind your back—figuratively, at least.
The bigger-picture problem is that existing privacy and civil rights safeguards developed slowly. They were mostly designed to limit old-fashioned forms of privacy violation, such as illegal searches or unauthorized revelation of private activities, such as medical treatments or bank transactions. New technologies such as facial recognition are like science fiction X-ray goggles. They can intrude in ways that, until recently, were impossible or impractical. While one facial recognition sensor might be minimally invasive, the use of many—widely and intensively—in a concerted manner, turbocharged by artificial intelligence, can comprehensively track a person’s movements, associations, and activities in real time, wherever they go. Such a powerful tool may cause a freedom-impinging violation far more profound than the sum of its individual intrusions. In so doing, the technology enables an end run around foundational principles and standards of law and ethics.
Proponents point out that facial recognition is effective for finding fugitives, as well as for recognizing unidentified crime victims and witnesses. But a cornerstone of our legal tradition, including the Bill of Rights, is that prosecutorial success should never be the only criterion for judging law enforcement tactics.
This year, San Francisco, Oakland, California and Somerville, Massachusetts banned law enforcement’s use of facial recognition. That will buy time to set policies before the technology becomes ubiquitous. But those bans don’t address use by private companies—or by locally-assigned federal agents. Portland City Council recently began discussing restrictions on the use of facial recognition technology. It would be great if it follows through, but officials will surely face intense, sustained pressure from tech firms, local businesses, and other interests to water down any such measure or delay it indefinitely.
A combination of factors will ensure the continued explosion of facial recognition, including accelerating technology, aggressive tech firm marketing, and security-related fearmongering. The way to protect rights and privacy is to actively press for strong local, state, and federal laws to control and manage its use. We cannot avoid dystopian facial recognition by burying our heads in the sand.