Despite its recent attention, the technology behind facial recognition has been in the works for over 50 years. In 1964, a team of scientists from the University of Texas started training a computer to recognize faces using mapping techniques. It was not until 2015 that facial recognition would become a widely known and available technology, thanks to social media apps like Snapchat and Facebook. We now use it to unlock our iPhones, sign into apps, and help the blind.

Privacy Concerns

Unsurprisingly, this technology has become increasingly controversial. Up until 2019, police would use technology that accessed license photos and mugshots to catch criminals. These photos were taken by, and for, the government, and the participant in front of the camera was aware they were being taken. However, in early 2020, an article by Kashmir Hill was published in the New York Times that detailed a new advancement in facial recognition tech: Scraping the entirety of the internet for photos to feed facial recognition AIs. Clearview AI’s technology scans millions of sites for photos including Facebook, Venmo, and YouTube for images. According to Hill, the three billion images that Clearview AI has captured “goes far beyond anything ever constructed by the United States Government or Silicon Valley giants.” The technology has been sold to police departments all over the country and possibly governments in other countries. According to Muckrock.com, a government watchdog site recognized by The Freedom of the Press Foundation, over 200 U.S. agencies already have access to this technology. This brings about serious questions regarding privacy and the use of data not intended for government purposes. Clearview claims that they only access images that are available to search engines. If a user sets their profile to be private to search engines, their photos won’t be in the data base. However, if the database already skimmed the photo, it’s too late for it to be removed. Founder, Ton-That, said there may be a way for users to request their photos be removed from the database in the future.

Clearview AI, interestingly enough, is backed by Peter Thiel, an outspoken Libertarian and funder behind the Hulk Hogan case that shut down Gawker.

“I’ve come to the conclusion that because information constantly increases, there’s never going to be privacy […] Laws have to determine what’s legal, but you can’t ban technology. Sure, that might lead to a dystopian future or something, but you can’t ban it.” — David Scalzo, Cleaview AI Backer

The rather blasé sentiment that Mr. Scalzo holds regarding privacy is concerning. He does make a point that we cannot ban technology. As we saw with the Betamax case, we cannot punish companies that provide advances in new technology for the missuses by its users. This could create a chilling effect in technological advances. However, The 4th Amendment gives citizens the right “to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures.” There is debate as to what is considered “public” on the internet. On one hand, there is no reasonable expectation of privacy for items available to search engines, such as public social media profiles. However, when that data is being stored, sold, and then utilized in criminal investigations, all without user knowledge or consent this opens up a different set of questions. Video surveillance in public places is also not against the 4th Amendment and passes the Katz test, as there is no expectation to privacy in public and the video surveillance is not sense-enhancing by definition.

Effects on the Black Community

In addition to privacy and data usage concerns, facial recognition has dangerous flaws baked into its programming that could cause human rights violations. A study by researcher, Joy Buolamwini, at the M.I.T. Media Lab’s Civic Media Group, found that facial recognition technology was highly inaccurate when trying to recognize non-white and non-male faces. The error rates for light-skinned men were less than one percent, while error rates for dark skinned women were between 20 and 34 percent. This is because the software is mainly fed photos of white, male faces according to Buolamwini. Considering black and brown people are already incarcerated at a higher rate than white people, the rate of accuracy of this technology could lead to an increase in racial profiling and false charges. We have seen this play out already. According to a June 24th report by Bobby Allyn at NPR, Robert Williams was arrested and detained for 30 hours for a crime he didn’t commit, thanks to facial recognition technology deployed by the Detroit Police Department. The lack of accuracy for black and brown people in combination with the interactions we’ve seen play out with the police, such as the George Floyd murder, is an incredibly dangerous mix.

During the Minneapolis protests over the George Floyd murder, Customs and Border Protection flew an unarmed Predator drone over protesters. Some police body cams have facial recognition software installed. In addition to police brutality and deadly clashes between the police and protestors, the addition of government aircraft and surveillance could have a chilling effect on protest, the right to which is a core value of our democracy.

In June of 2020, Amazon, creator of widely used Rekognition, and Microsoft both publicly stated they would place moratoriums on selling facial recognition software until there are national laws “grounded in human rights” regarding the software. A multi-faceted solution is warranted in situations involving technology. In addition to clear, federal government regulation that protects the privacy of citizens, technology firms should also be self-governing. As those who understand the technology and it’s implications the best, Tech, in cooperation with government should pass laws protecting the public. Congress passed the Stored Communications Act in 1986 to attempt to provide privacy to citizens in regards to the internet, however these terms need updated. When companies use other peoples photos without consent and use them for financial gain and public surveillance, the affected should be made aware and given an option to consent. In 2019, Senator Coons and Senator Lee brought a bill before congress to address this. The Facial Recognition Technology Warrant Act of 2019 attempts to both allow police to continue to use facial recognition if they have warrants, while protecting constituents. Laws like these, in combination with the support of the tech community, could help reform facial recognition tech and it’s uses.

Facial Recognition: The 4th Amendment and Police Relations was originally published in All Things on Medium, where people are continuing the conversation by highlighting and responding to this story.