(Ed. Note: The “Innovation Fact of the Week” appears as a regular feature in each edition of ITIF’s weekly email newsletter. Sign up today.)
The United States faces an overdue reckoning for a pattern of systemic racial injustice and police brutality following the recent shootings of George Floyd, Breonna Taylor, Rayshard Brooks, and countless other Black Americans. These events have prompted some technology companies to announce last week that they would not sell facial recognition technology to U.S. law enforcement agencies until Congress passes a federal law establishing sufficient guardrails on the use of the technology. But some privacy advocates have said rules are not enough and Congress should ban police from using the technology altogether.
Regardless of where one stands on police reform, banning facial recognition for law enforcement makes little sense. To understand the reasons, consider some of the most popular ideas for responding to police violence.
One major current proposal is to “defund the police” and redirect these funds to invest in the community with better healthcare, education, and housing. For some, this means abolishing entire police departments that they believe are inherently racist and irredeemable. In this case, banning facial recognition from law enforcement would be moot since there would be no police agency to speak of.
For others, “defund the police” means scaling back funding for police departments to significantly cut the number of officers on the streets and thereby reduce incidents of police violence. In this case, using facial recognition could be part of the solution as it automates investigative tasks that police officers would otherwise have to do manually. If police are more productive, communities can hire fewer of them. Banning facial recognition technology would undercut efforts to reduce the size of police departments.
Another major proposal is to implement significant reforms to establish better standards (e.g., prohibiting chokeholds and no-knock warrants), better accountability (e.g., ending qualified immunity), and more transparency (e.g., requiring bodycams and dashcams). In this scenario, bans would be unnecessary because policymakers would have the opportunity to establish better standards, accountability, and transparency for police use of facial recognition technology as well. Ideally, this would include requirements that law enforcement agencies only use facial recognition technology that meets high accuracy performance standards; that agencies disclose their use of the technology, the data sources for their images, and their data retention policies; and that agencies establish controls to limit inappropriate use of the technology, such as warrantless surveillance of political events. Banning facial recognition would also make it harder for police agencies to quickly disclose bodycam and dashcam video to the public because the software used to obscure faces uses this technology.
Some have called for more targeted reforms, such as increasing implicit bias training for police officers and increasing the diversity of police workforces, based on the assumption that police agencies are generally good but need to provide their officers better training and exposure to their own prejudices. In this case, using facial recognition technology would be useful at reducing bias because it would reduce instances where individual officers misidentify people. Studies have shown that humans are often bad at identifying faces, particularly when they are looking at people of a different race than themselves. Using facial recognition technology can mitigate some of these human biases, as the best algorithms are much more accurate than humans and show “undetectable” differences between different demographics.
Finally, some say this technology should be banned until Congress makes reforms. But these critics miss the fact that the benefits of law enforcement use of facial recognition are well-proven—they are used today to help solve crimes, identify victims, and find witnesses—and most of the concerns about the technology remain hypothetical. In fact, critics of the technology almost always make a “slippery slope” argument about the potential threat of expanding police surveillance, rather than pointing to specific instances of harm. Banning the technology now would do more harm than good.
Calls to ban facial recognition are misguided and may even undercut reforms. It’s unfortunate that self-described privacy advocates have tried to make the Black Lives Matter movement about their own pet issue. It’s not, and it shouldn’t be. Addressing police violence and systemic racism is an urgent and important issue, and policymakers should remain firmly focused on that task.