Law enforcement agencies and industry lobbyists are starting to push back against some of the facial recognition bans that were put in place across the United States. A little more than two dozen state and local governments have implemented facial recognition laws in the past few years, beginning in 2019 with a public facial recognition ban in San Francisco.
In most cases, those bans reflect privacy and accuracy concerns about facial recognition technology. Surveillance tech can intrude on people’s private lives, and many facial recognition solutions were demonstrably less accurate when applied to people with darker skin.
However, those considerations are now being weighed against increasing crime levels across the country. New Orleans lawmakers are considering a repeal after reports found that homicide is up 67 percent in the city in the past two years, and California is reconsidering its 2019 ban on facial recognition in police body cameras in light of increasing rates of theft. The ban is set to expire on January 1, and would need to be renewed to remain in effect.
As it stands, Virginia has already repealed its blanket facial recognition ban, though it has replaced it with a new policy that includes certain safeguards for those using facial recognition during investigations. Most notably, Virginia’s new policy still bans the use of facial recognition in live surveillance systems, and states that any facial recognition system that the police want to use need to be at least 98 percent accurate across demographics. It also states that a facial recognition match is not sufficient cause for a warrant or an arrest.
Those safeguards could eventually serve as a template for facial recognition laws in other jurisdictions. Privacy advocates, meanwhile, are still calling for more meaningful protections, though they acknowledged that Virginia’s policy is better than no policy at all. As for potential improvements, privacy advocates suggested that the police should need to obtain a warrant in order to perform a facial recognition search.
It is worth noting that the amount of bias in facial recognition systems has decreased in the past few years. In its most recent Biometric Technology Rally, the Department of Homeland Security found that the top facial recognition solutions achieved similar accuracy rates regardless of skin color, though not every system is able to perform to that standard. It is also possible for biased police departments to use non-biased technology in a discriminatory fashion, through an uneven distribution of cameras or through targeted policing practices.
In all likelihood, lawmakers will start to move away from blanket facial recognition bans, and instead implement laws that allow for the use of the technology in certain circumstances. The challenge will be finding a balance that protects people’s civil liberties while still helping investigations to proceed.
Until then, activists will face sterner opposition as they continue to push for facial recognition legislation. That opposition will come from the police and from developers like Clearview AI, who have upped their lobbying efforts to create a police market for their products.
May 18, 2022 – by Eric Weiss