Facial recognition technology, once a darling of Silicon Valley with applications for policing, spying and authenticating identities, is suddenly under fire. Conservative Republicans like Representative Jim Jordan of Ohio and liberal Democrats like Representative Alexandria Ocasio-Cortez of New York have strongly criticized the technology. San Francisco, Oakland, Berkeley and Somerville, Mass., have barred all of their government agencies, including the police, from using it. And several Democratic candidates for president have raised deep concerns about it, with one, Senator Bernie Sanders, calling for an outright ban for policing.
We think the senator is right: Stopping this technology from being procured — and its attendant databases from being created — is necessary for protecting civil rights and privacy. But limiting government procurement won’t be enough. We must ban facial recognition in both public and private sectors, before we grow so dependent on it that we accept its inevitable harms as necessary for “progress.” Perhaps over time appropriate policies can be enacted that justify lifting a ban. But we doubt it.
The essential and unavoidable risks of deploying these tools are becoming apparent. A majority of Americans have functionally been put in a perpetual police lineup simply for getting a driver’s license: Their D.M.V. images are turned into faceprints for government tracking with few limits. Immigration and Customs Enforcement officials are using facial recognition technology to scan state driver’s license databases without citizens’ knowing. Detroit aspires to use facial recognition for round-the-clock monitoring. Americans are losing due-process protections, and even law-abiding citizens cannot confidently engage in free association, free movement and free speech without fear of being tracked.
Yet industry officials, lawmakers and even some privacy advocates have been skeptical about banning the technology, as we and others proposed more than a year ago. Other than clearly malicious technologies like spyware, the United States has banned very few technologies, even those used in warfare. Those making a case against a ban on facial recognition muster three main arguments.
The first argument is that the benefits of using a new technology can often outweigh the harms. Law enforcement officials say that facial recognition helps catch criminals, find missing people and prevent crimes. Industry officials point to the convenience of being recognized by your phone, in photos and as you’re boarding a plane.
Instead of proposing a ban, these advocates of regulating facial recognition with a light touch assert that lawmakers should create laws that incentivize law enforcement to address systemic bias in their procedures and require companies and government agencies to adopt rules that heighten transparency and accountability. A guiding hope is that requiring consent forms or search warrants before facial recognition technology is applied to our images will protect our rights.
We disagree. “Notice and choice” has been an abysmal failure. Social media companies, airlines and retailers overhype the short-term benefits of facial recognition while using unreadable privacy policies and vague disclaimers that make it hard to understand how the technology endangers users’ privacy and freedom.
And while warrant requirements are important for limiting the power of government officials, they don’t apply to the private sector, where individuals and companies will remain largely free to monitor their families, neighbors, co-workers and rival businesses as they choose, including by sharing the information with law enforcement.
A second argument for little or no regulation of facial recognition is that strong fears about new technologies are overreactions. Apostles of innovation compare people who are calling for banning facial recognition to the naysayers of yesteryear whose anxieties about new technologies ranging from the automobile to photography proved unfounded. From this perspective, facial recognition technology is the new fingerprint.
But the tangible harms of facial recognition are potentially far more menacing. The technology is less accurate with people of color and is biased along gender lines. And things will not get better as it becomes more accurate, because big companies, government agencies and even your next-door neighbors will seek to deploy it in more places. They will want to identify and track you. They will want to categorize your emotions and identity. They will want to infer where you might shop, protest or work — and use that information to control and manipulate you, or deprive you of opportunities.
It is likely that the technology will be used to police social norms. People who skip church or jaywalk will be noticed — and potentially ostracized. And you’d better start practicing your most convincing facial expressions. Otherwise, during your next job interview, a computer could code you as a liar or malcontent.
A third argument in favor of using facial recognition technology is that privacy and civil liberties are best protected by creating rules that focus on all surveillance, not just a particular technology. If we ban facial recognition today, this argument goes, what happens in the future when gait recognition or devices that read brain patterns go mainstream? Some argue that banning this technology or some other one will prevent us from wrestling with larger questions that apply to all emerging technologies, like whether there is a right to be anonymous in public. Many of those questions are likely to be taken up in the courts.
But we believe society can’t wait years for institutions like the Supreme Court to update privacy protections for the digital age. By then, facial recognition infrastructure will be ubiquitous, and exploiting its full potential will seem like a good use of resources. The law singles out specific technologies all the time because they are so exceptional. Automobiles, spyware, medical devices and a host of other technologies have their own specific rules. Airplanes and telecommunications technologies were given their own federal regulatory agencies.
Facial recognition is truly a one-of-a-kind technology — and we should treat it as such. Our faces are central to our identities, online and off, and they are difficult to hide. People look to our faces for insight into our innermost feelings and dispositions. Our faces are also easier to capture than biometrics like fingerprints and DNA, which require physical contact or samples. And facial recognition technology is easy to use and accessible, ready to plug into police body cameras and other systems.
We support a wide-ranging ban on this powerful technology. But even limited prohibitions on its use in police body cams, D.M.V. databases, public housing and schools would be an important start.
The public is ready for this, and the actions by San Francisco, Somerville, Berkeley and Oakland show it: Our society does not have to allow the spread of new technology that endangers our privacy.
Evan Selinger and Woodrow Hartzog.