ACLU Uses Amazon’s Rekognition to Match Members of Congress with Mugshots

“…in responding the ACLU’s latest volley, a spokesperson emphasized that Rekognition is used to aid human review, and not to definitively link one facial image to another”

The ACLU has turned Amazon’s facial recognition technology against the company itself in a damning new demonstration highlighting the technology’s flaws.

The civil rights organization says that it used Amazon’s face matching Rekognition service to see what would happen when it compared the images of members of Congress against a public database of 25,000 mugshots. It got 28 matches. Worse, 39 percent of the matches were of people of color, even though this group only comprises 20 percent of Congress, validating the ACLU’s longstanding criticisms of facial recognition technologies over their potential for racial bias.

In a blog post announcing the findings, the ACLU said it “used the default match settings that Amazon sets for Rekognition.” And the organizations noted that the exercise only cost it $12.33, underlining the large-scale potential for abuse now that such technology is cheaply available.

The ACLU’s announcement comes in the wake of intense criticism over Amazon’s sale of facial recognition technology to government and police agencies, with some of the company’s own shareholders and employees adding their voices to the chorus. And this reflects a broader dialectic over the rights and ethical implications of biometric surveillance technologies as their public use continues to escalate.

For its part, Amazon insists that its facial recognition technology is a tool that offers real benefits to society, such as by helping authorities to track down missing children. At the same time, in responding the ACLU’s latest volley, a spokesperson emphasized that Rekognition is used to aid human review, and not to definitively link one facial image to another. In other words, Amazon says it’s a helpful tool that isn’t highly accurate.

Sources: ACLU, Reuters

–

July 27, 2018 – by Alex Perala