Legislative action at the municipal, state, and federal levels is accelerating on the issue of whether facial recognition technology, and the proliferating use of it by law enforcement agencies across the country, represents an unconstitutional violation of individual rights.
On June 4, the U.S. House of Representatives Committee on Oversight and Reform held the second of three planned hearings on the potential dangers posed to Americans’ constitutional liberties and privacy rights when facial recognition is used improperly, or without adequate oversight, by government agencies. Happily—and strikingly given the degree of political polarization rampant today—one of the most important things to emerge during this second hearing was the widespread bipartisan consensus that this technology needs to be brought under stricter control to protect civil liberties. Where else would the very liberal chair of the committee, Rep. Elijah Cummings (D-MD) and the very conservative chair of the so-called Freedom Caucus, Mark Meadows (R-NC), agree on much of anything?
This federal effort to identify the possible misuses of this technology coincides with recent state and local conversations echoing similar concerns. San Francisco passed an absolute ban on the use of facial recognition by the police and other government authorities, and similar bans are being considered by Oakland, CA and Somerville, MA. Democratic state legislators in Massachusetts and California have introduced bills banning facial recognition, or at use of it by the police, at a statewide level. And Senator Roy Blunt (R-MO) introduced a bill banning uses of facial recognition in certain commercial contexts.
Facial recognition technology, as a category, is best described as any system that algorithmically and automatically “analyze[s] video, photos, thermal captures, or other imaging inputs to identify or verify a unique individual,” according to testimony provided to the Committee by Joy Buolamwini, a researcher at the Massachusetts Institute of Technology. The negative implications are easy to grasp: the misuse of this technology could inhibit or chill the free expression rights of Americans who may decide not to show their support for one cause or another if they believe their attendance at a meeting or speech is being recorded by the government.
Neema Singh Giuliani of the American Civil Liberties Union testified that the government could acquire the kind of “sensitive information” about an individual, including with whom they associate or whether they attend a political rally, that Supreme Court precedent suggests must only be obtained with a warrant. However, the ACLU states that, “in most cases,” the government use of facial recognition lacks “judicial scrutiny of any kind.” And this abnegation of due process does not occur only occasionally or in scattered circumstances. Indeed, the utility of the technology actually depends on mass collection of information about many people. The ACLU estimates that “[f]ederal, state, and local law enforcement agencies have used face recognition hundreds of thousands of times,” in clear violation of Americans’ constitutional rights. And while there haven’t been any definitive rulings on this emerging set of technologies, several cases—including a landmark judgment that law enforcement agencies need a warrant to make use of thermal imaging technology—indicate a legal path forward for more oversight.
But facial recognition technology is not only potentially infringing on our constitutional protections against “unreasonable searches and seizures” (Fourth Amendment), and chilling people’s exercise of their freedom to express themselves (First Amendment). It also runs afoul of the equal protection clause of the Fourteenth Amendment in breathtaking fashion when it fails to work as intended. And fail it does: the ACLU notes that facial recognition “is disproportionately inaccurate on certain subgroups,” including individuals with darker skin pigmentation, meaning that over-policed communities of color could be subjected to the additional burden of misidentification. When the ACLU tested official congressional portraits against a facial recognition algorithm, 40 percent of the false matches were Representatives of color, even though just over 20 percent of congressional seats are held by people of color.
Already widespread, the use of facial recognition continues to grow. The Georgetown Center for Privacy and Technology testified to the House committee that more than half of Americans have already had their identities recorded in one of many extant facial recognition databases. Police forces in Chicago and Detroit already have the capacity to use facial recognition, while New York City, Orlando, and Washington, DC are all exploring pilot programs to “identify people in real-time, remotely and in secret, from video feeds that keep a constant eye on these cities’ streets,” according to the Center for Privacy and Technology. Sure, providing law enforcement with effective tools to identify and convict perpetrators of crime in a changing technological landscape is a worthwhile goal, especially with state and local budgets groaning under strain. But we ought not let these agencies, however badly stretched they may be, trample over our most basic constitutional protections through the unrestricted use of powerful new tools.
The House Committee on Oversight and Reform is doing important work. The investigation itself is important as an awareness-raising measure, but it is also trying to push through meaningful changes. For example, the Committee has demanded that the Federal Bureau of Investigation return for another hearing to confirm that it is implementing the recommendations made earlier this year by the Government Accountability Office regarding the appropriate use of facial recognition. At the time of the latest hearing on June 4, the FBI had fully implemented only one of six recommendations.
Representative Carol Miller (R-WV) explicitly drew a connection between the use of facial recognition in the United States and its use in China. The ruling Chinese Communist Party uses facial recognition as part of a widespread and growing surveillance system often deployed to suppress political dissent. Mrs. Miller stated that her recent trip to China provided “a stark contrast to what our privacy and our liberty is in America” and “poses many questions to the United States” about facial recognition’s “appropriate use.” We’re not at Beijing levels yet, not by a long shot. But the possibility of stumbling into the terrifying reality that’s emerging in China, even by mistake, should give us plenty of pause, and give momentum for reform. If Republicans and Democrats can work together to rein in this emerging danger, then that may also open up possibilities for cooperation on other fronts as well. And that would be a good thing, too.