قالب وردپرس درنا توس
Home / Technology / Face recognition software incorrectly identifies 28 legislators as crime suspects: NPR

Face recognition software incorrectly identifies 28 legislators as crime suspects: NPR



The American Civil Liberties Union says that Amazon Recognition, an online facial recognition software that inaccurately identifies lawmakers and threatens civil rights ̵

1; charges that Amazon denies.

Leon Neal / AFP / Getty Images


hide caption

toggle caption

Leon Neal / AFP / Getty Images

The American Civil Liberties Union says that Amazon Recognition, an online facial recognition software that inexactly identifies lawmakers and threatens civil rights – charges that Amazon denies.

Leon Neal / AFP / Getty Images

Amazon's facial recognition software mistakenly identified 28 congressmen as being arrested for crimes, the American Civil Liberties Union announced Thursday.

Amazon Rekognition was marketed as a tool that allows extremely accurate facial analysis photos and video.

The ACLU tested this claim using the software to scan photos of each current member of the House of Representatives and the Senate in a database created by the watchdog from thousands of publicly available arrest photos.

"Members of Congress who were falsely compared to the Mugshot database we used in the test are Republicans and Democrats, men and women, and legislators from all over the country." the ACLU detected.

The test identified high-color people – 39 percent – though they make up only 20 percent of the congress. One member falsely cited as a crime suspect was MP John Lewis, D-Ga., Who first became known as a civil rights leader.

As part of the test, the ACLU said they had used Amazon's default match settings. 19659008] But a spokeswoman for Amazon Web Services said in an email message that the ACLU should have changed those settings – using a higher "threshold" or percentage indicating how confident Rekognition is of finding a match ,

"While 80% confidence is an acceptable threshold for photos of hot dogs, chairs, animals or other social media use cases, it would not be appropriate to identify individuals with a reasonable level of safety," she said. For law enforcement agencies, Amazon leads customers to set the threshold at least 95 percent.

North Californian lawyer Jacob Snow's ACLU responded to this comment in an e-mail message: "We know from our test that Amazon is not making any effort asking users what they're using recognition for," he said.

Snow does not believe that changing the threshold changes the danger: "The technology of facial surveillance in the hands of the government is prepared for abuse and raises serious civil rights concerns."

The outcry of privacy and civil rights groups has prosecuted not deterred from pursuing the technology. Orlando, Florida police tested Rekognition's real-time monitoring. The Washington County Sheriff's Office, near Portland, Oregon, has used it to search faces of photos of suspects taken by MPs.

"This is partly a result of salespeople's face recognition technology because it becomes another source of revenue," said Jeramie Scott, national security consultant at the Electronic Privacy Information Center in Washington, DC. opposite NPR. He compared facial recognition software with body cameras worn by law enforcement agencies, which can be used for police accountability or increasingly for public monitoring.

He stressed the need for debate so that technology will not be a bad solution for bad politics. "Because of the disproportionate error rate and the real risk of depriving us of the civil liberties inherent in facial recognition technology, we need to talk about how, when, and under what circumstances this technology should be used by law enforcement, if at all." [19659021]
Source link