A growing backlash against face recognition suggests the technology has a reached a crucial tipping point, as battles over its use are erupting on numerous fronts. Face-tracking cameras have been trialled in public by at least three UK police forces in the last four years. A court case against one force, South Wales Police, began earlier this week, backed by human rights group Liberty. Ed Bridges, an office worker from Cardiff whose image was captured during a test in 2017, says the technology is an unlawful violation of privacy, an accusation the police force denies. Avoiding the camera's gaze has got others in trouble.
Why the American Civil Liberties Union is calling out Amazon's facial recognition tool, and what the ACLU found when it compared photos of members of Congress to public arrest photos. A group of Amazon shareholders is pushing the tech giant to stop selling its controversial facial recognition technology to U.S. government agencies, just days after a coalition of 85 human rights, faith, and racial justice groups demanded in an open letter that Jeff Bezos' company stop marketing surveillance technology to the feds. Over the last year, the "Rekognition" technology, which has been reportedly marketed to the U.S. Immigration and Customs Enforcement (ICE), has come under fire from immigrants' rights groups and privacy advocates who argue that it can be misused and ultimately lead to racially biased outcomes. A test of the technology by the American Civil Liberties Union (ACLU) showed that 28 members of Congress, mostly people of color, were incorrectly identified as police suspects. According to media reports and the ACLU, Amazon has already sold or marketed "Rekognition" to law enforcement agencies in three states.
Let's start with some comments about a recent ACLU blog in which they run a facial recognition trial. Using Rekognition, the ACLU built a face database using 25,000 publicly available arrest photos and then performed facial similarity searches of that database using public photos of all current members of Congress. They found 28 incorrect matches out of 535, using an 80% confidence level; this is a 5% misidentification (sometimes called'false positive') rate and a 95% accuracy rate. The ACLU has not published its data set, methodology, or results in detail, so we can only go on what they've publicly said. To illustrate the impact of confidence threshold on false positives, we ran a test where we created a face collection using a dataset of over 850,000 faces commonly used in academia.
When is it appropriate for police to conduct a face recognition search? To figure out who's who in a crowd of protesters? To monitor foot traffic in a high-crime neighborhood? To confirm the identity of a suspect -- or a witness -- caught on tape? According to a new report by Georgetown Law's Center on Privacy & Technology, these are questions very few police departments asked before widely deploying face recognition systems.
Future Tense is a partnership of Slate, New America, and Arizona State University that examines emerging technologies, public policy, and society. The ACLU released a report on Thursday revealing that Rekognition, Amazon's facial recognition tool, had falsely matched 28 members of Congress to mug shots. Members of the ACLU purchased the version of Rekognition that Amazon offers to the general public and ran public photos of every member of the House and Senate against a database of 25,000 arrest photos. The entire experiment costed $12.33, which, as ACLU attorney Jake Snow writes in a blogpost, is "less than a large pizza." Almost 40 percent of the representatives that Rekognition falsely matched were people of color, even though they make up only 20 percent of Congress.