News

Actions

Amazon's face recognition technology falsely matched 28 members of Congress to crime suspects

Amazon's face recognition technology falsely matched 28 members of Congress to crime suspects
Posted at 3:45 AM, Aug 01, 2018
and last updated 2018-07-31 23:51:21-04

(RNN) - Amazon’s new facial recognition software identified 28 members of Congress as criminal suspects, according to the American Civil Liberties Union.

The ACLU said that faces from both men and women of all ages were tested in the new program called Rekognition.

Most of the faces misidentified by the technology were those who are black and Latino, according to the test study results released July 26.

“This test confirms that facial recognition is flawed, biased and dangerous,” said Jake Snow, an ACLU attorney.

Six of the misidentified legislators were members of the Congressional Black Caucus, and three of them wrote a letter to Amazon CEO Jeff Bezos.

According to the caucus’ website, the letters expressed their concern over the Rekognition technology, in wake of recent police brutality cases involving people of color.

Amazon has touted that its Rekognition software provides highly accurate facial analysis.

According to the New York Times, Nina Lindsey, an Amazon Web Services spokeswoman, said in a statement that the company’s customers had used its facial recognition technology for various beneficial purposes, including preventing human trafficking and reuniting missing children with their families. She added that the ACLU had used the company’s face-matching technology differently during its test than the company recommended for law enforcement customers.

“It is worth noting that in real-world scenarios, Amazon Rekognition is almost exclusively used to help narrow the field and allow humans to expeditiously review and consider options using their judgment,” said Lindsey. "The test results could have been improved by increasing the confidence threshold. While 80 percent is an acceptable threshold for photos of everyday items and objects, it's not appropriate for identifying individuals with a reasonable level of certainty."

A full list of the 28 members of Congress who were falsely matched can be found on ACLU.org.

Copyright 2018 Raycom News Network. All rights reserved.