On June 28, 2018, a man shot and killed f ive people at the Annapolis, Maryland newspaper the Capital Gazette. After law enforcement apprehended the suspect on the scene, they say, they used facial recognition technology to identify him as Jarrod Ramos. The cops made a big show of their use of the technology to identify the shooter, repeating this part of the story over and over to any reporter who’d listen. Probably not coincidentally, these statements were issued just weeks after the ACLU made national news by calling on Amazon to stop selling its face surveillance product, Rekognition, to law enforcement. The message from law enforcement appeared to be: Don’t worry about the ACLU’s alarmist chatter on face surveillance; it’s a necessary public safety tool, and whether the ACLU likes it or not, we’re going to use it.
But is it a necessary tool? In the case of the Capital Gazette shooting, face surveillance technology did not save any lives. The suspect was already in police custody when they reportedly used face recognition technology to identify him. The use of the tool in this case did not protect anyone. And contrary to claims we’ll hear for the next few decades from law enforcement about the life saving properties of face surveillance tools, it’s unlikely that the technology will ever stop a mass shooting, terrorist attack, or other serious threat to public safety.
And that’s for a very simple reason: Face surveillance tools may be good, in some cases, at putting names to faces, but the technology does not and will not magically tell law enforcement who is dangerous and who is not.
Mass shootings, as a rule, occur without warning. No one at the Mother Emanuel AME Church in Charleston knew Dylann Roof was going to kill people before he walked through the church’s doors in June 2015 and opened fire on worshippers. Security at the Mandalay Bay hotel were not told to look out for Stephen Paddock, the man who allegedly set up a sniper’s den in his hotel room and shot hundreds of concertgoers assembled outside in 2017. Had the Sandy Hook elementary school used face recognition to identify visitors to its campus, that technology would not have identified Adam Lanza as a threat, because prior to his attack, no one knew he was going to kill children that day.
Face surveillance technology cannot predict the future. It is foolish to expect that widespread adoption of the technology will protect us from these kinds of incidents, particularly when the technology has so many problems. Studies have shown the tool doesn’t recognize dark skinned women as well as it IDs white men. It doesn’t work well in low-light conditions, or at a severe angle—the very type of angle most surveillance cameras provide, from their perches on the sides of buildings and the tops of light poles. And according to a new video from the Wall Street Journal, technologies in use by entities like schools are so easy to game that someone can literally print out a photograph of someone else, show it to the camera, and gain entry to a space that’s supposedly secured by face surveillance.
Please note that by playing this clip you are accepting cookies from YouTube and Google.
Too often, when law enforcement and private corporations tell us a surveillance technology will protect public safety, many people simply accept that claim as fact, and move on to discussions about how we can best integrate the tool without sacrificing too much of our privacy or liberty. But we shouldn’t accept these claims without evidence. And right now, there is simply no evidence for the claim that the use of widespread face surveillance will protect the public from the most serious threats to our collective safety: shootings and other incidents of mass violence.
The stakes here are extremely high. As privacy scholars have meticulously argued , face surveillance technology is different from—and more dangerous than—all other existing forms of surveillance. If it is adopted in a widespread manner, the tool quite literally has the potential to wipe out privacy and anonymity in public, rendering obsolete our rights to freedom of speech, association, and assembly. It is anathema to freedom and liberty, which is why authoritarian societies like China are making quick use of it to keep their subjects in line. While it isn’t good at stopping mass violence, it is a near perfect tool for social and political control. Far be it from protecting public safety, the widespread adoption of face surveillance by law enforcement in the US is likely to exacerbate harms against the typical targets of police surveillance and harassment: Black and Latinx people, immigrants, dissidents, and the poor.
The next time you hear law enforcement make big claims about how face surveillance technology is necessary to protect public safety, think critically about those claims. Ask yourself: How could this technology actually protect me? And equally important, ask yourself: if law enforcement adopts this technology across the country, using it to scan and identify the faces of every person who passes by the millions of surveillance cameras across the United States, what will we lose? I think you’ll find that the cost-benefit analysis doesn’t weigh in the favor of face scanning technology, and that it won’t be a close call.