Despite facial recognition's seal of approval from law enforcement agencies across the U.S., recent experiments show the technology is far from infallible. In a demonstration by the American Civil Liberties Union, about 26 California lawmakers were misidentified by face-matching software built by Amazon, putting the rate of a mismatch at about 1 in 5. The results mimic a similar test done by the advocacy group in 2018 when a test saw Amazon's software, called'Rekognition', mismatch 28 members of congress -- many of whom were people of color. The ACLU says a test of Amazon's facial recognition software misidentified 1 in 5 lawmakers fed into its system Similarly, the software attempted to match their head shots against a database of known criminals -- a process that has become commonplace for the at least 200 departments across the U.S. who use Rekognition software. According to the LA Times, the test is fueling calls from California legislators to limit the technology's application in a law enforcement capacity, including its integration with police body cameras.
Aug-13-2019, 19:30:22 GMT