In May, Amazon came under fire for working with law enforcement departments across the US to deploy facial recognition technology, appropriately named "Rekognition." Information on the tech was publicly viewable through Amazon's Web Services website, but it flew under the public's radar for quite some time.
As we mentioned at the time, Rekognition possesses the ability to track up to 100 individuals in any given image, pulling data from a library containing "tens of millions" of faces. Shortly after the news broke, one of Amazon's Rekognition partners -- Orlando's Police Department -- dropped out of the partnership, acknowledging the dangers of such tech.
Now, it seems those dangers have become evident, in a slightly embarrassing manner. The ACLU recently tested Amazon's Rekognition tech, only to find that it matched 28 US lawmakers with individuals arrested for a crime. While their exact testing methodology isn't known, the organization claims it performed its experiments with "80 percent" accuracy.
Worse yet, these tests were incredibly cheap to run. According to the ACLU, the entire procedure only cost about $12.33.
The ACLU also claims people of color were disproportionately matched with arrested men and women, a trend the civil rights group believes is evidence of racial bias.
Amazon has since taken issue with the ACLU's report, though. "While 80 percent confidence is an acceptable threshold for photos of hot dogs, chairs, animals and other social media use cases, it wouldn't be appropriate for identifying individuals with a reasonable level of certainty," the company said in a statement to The Verge.
Regardless of whether or not the ACLU's testing procedures were up to par, the bottom line here is that the technology still appears to be in its early stages. As such, deploying it for law enforcement purposes could be a dangerous path for Amazon to pursue - only time will tell whether or not the company will stand firm in the face of public pressure.