A coalition of civil rights groups, in a letter released Tuesday, called on Amazon to stop selling the program to law enforcement because it could lead to the expansion of surveillance of vulnerable communities.
“We demand that Amazon stop powering a government surveillance infrastructure that poses a grave threat to customers and communities across the country,” the groups wrote in the letter.
Amazon spokeswoman Nina Lindsey did not directly address the concerns of civil rights groups. “Amazon requires that customers comply with the law and be responsible when they use AWS services,” she said, referring to Amazon Web Services, the company’s cloud software division that houses the facial recognition program. “When we find that AWS services are being abused by a customer, we suspend that customer’s right to use our services.”
She said that the technology has many useful purposes, including finding abducted people. Amusement parks have used it to locate lost children. During the royal wedding this past weekend, clients used Rekognition to identify wedding attendees, she said. (Amazon founder Jeffrey P. Bezos is the owner of The Washington Post.)
The details about Amazon’s program illustrate the proliferation of cutting-edge technologies deep into American society — often without public vetting or debate. Axon, the maker of Taser electroshock weapons and the wearable body cameras for police, has voiced interest in pursuing face recognition for its body-worn cameras, prompting a similar backlash from civil rights groups. Hundreds of Google employees protested last month to demand that the company stop providing artificial intelligence to the Pentagon to help analyze drone footage.
The technology works through pattern recognition: Customers put known images – of child pornography or of celebrities, for example – into a database, and the software uses artificial intelligence to scan new images for a match with those already stored. The more images that are fed into the system, the more accurate the software becomes.
Amazon publicly introduced Rekognition in November 2016. Marketers could use the image recognition software, developed by the company’s scientists to analyze billions of images and videos daily, to recognize celebrities in their videos, while owners of dating apps could use the program to identify unwanted suggestive or explicit content, according to the company’s website.
But the vast expansion of facial recognition and other image-scanning technology has been highly controversial and given rise to privacy concerns. Facial recognition can allow strangers to identify people who don’t wish to be identified, such as shoppers in stores, individuals in a crowd, or people who appear in photos that get posted on social media.
Privacy advocates point out that in an era in which everyone has a camera on their smartphone, cities have put cameras on traffic stops and other public infrastructure, and police are wearing body cameras, the opportunity to have one’s photograph taken, identified, analyzed, and stored in perpetuity has grown immensely.
“Once powerful surveillance systems like these are built and deployed, the harm can’t be undone. We’re talking about a technology that will supercharge surveillance in our communities,” said Nicole Ozer, Technology and Civil Liberties Director for the ACLU of Northern California. She said the technology could be used “to track protesters, target immigrants, and spy on entire neighborhoods.”
The documents provide a detailed look at how Amazon is marketing Rekognition. It can identify up to 100 people in a crowd, the documents said.
The sheriff’s office of Washington County, Ore., built a database of 300,000 mug shots of suspected criminals that officers could have Rekognition scan against footage of potential suspects in real-time. The footage could come from police body cameras and public and private cameras. The county pays Amazon between $6 and $12 a month for the service, a county spokesman said.
According to the documents, Amazon asked the county to tout its experience with Rekognition to other public sector customers, including a manufacturer of body cameras.
Deputy Jeff Talbot, public information officer for the Washington County Sheriff’s Office, said the program was not operating in the shadows and had been the subject of several news local stories. He pointed out that jail booking photos are already public and that the software simply allows officers to scan them instantaneously and in real-time, and compare them against footage of actual suspects, which is a valuable contribution to public safety. “Our goal is to inform the public about the work we’re doing to solve crimes. It is not mass surveillance or untargeted surveillance.”
He could not say how many crimes the program had helped solve and added that the software wasn’t always accurate. But he said officers were trained not to rely exclusively on the software to make decisions, and it was just an additional tool in the officer’s tool kit. For the cheap price Amazon was offering, he said it made sense to test out the service.
Zahra Billoo, executive director of the Council on American-Islamic Relations San Francisco Bay area office, one of the groups that signed the letter, said many people who are booked into jail are not always charged with a crime or are proved innocent. She said that she worried that people’s civil rights are violated when law enforcement keeps their images in a database even after they are proved innocent or were never charged. She said Amazon was contributing to these violations by making it easier to scan people’s faces, repeatedly subjecting them to surveillance.
In addition to the ACLU, the coalition of about 40 groups included Color of Change, Human Rights Watch, Muslim Advocates and the Electronic Frontier Foundation.
Amazon is one of many companies selling artificial intelligence tools such as facial recognition and image-scanning to business clients. Microsoft offers a rival service, called Facial Recognition API. A crop of start-ups market the ability to scan the emotions on people’s faces as they walk in and out of stores. Such technology has been touted as a way to prevent shoplifting.