AWS Facial Recognition Platform Misidentified Over 100 Politicians As Criminals
Facial recognition technology continues to present privacy and security issues as it misidentifies people at an alarming rate. The technology is used by law enforcement agencies to make arrests, despite these identification errors. Consumer privacy expert Paul Bischoff found that Amazon’s face recognition platform mistook more than 100 photos of US and UK lawmakers, labeling them as wanted criminals.
The technology produced by Amazon, Rekognition, is a could-based facial recognition platform launched in 2016 and is used by several different US agencies, including ICE in Orlando and the Florida Police. Bischoff also found that the platform was racially biased, and misidentified non-white people at a higher rate than white people. Bischoff found that Rekognition misidentified four more individuals during his experiment than one conducted by the American Civil Liberties Union two years ago.