Amazon's Rekognition Falsely Identified Congress Members In Mugshot Database

Amazon has been under some heat lately for its facial recognition tool dubbed 'Rekognition'. This tool was used by some police forces to help catch some criminals. But it has come under fire lately for not being all that accurate. The ACLU or American Civil Liberties Union, did an experiment to see how well this tool actually worked. And Rekognition falsely identified 28 members of Congress - from both the House and Senate - in a mugshot database. Showing one of the worst failure rates imaginable.

In the experiment, the ACLU found that Rekognition had the worst failure rate for people of color. The ACLU had built a database of around 25,000 pubclicly available mugshots, and then used Rekognition to search the database for each member of the Senate and House of Representatives. There were 28 false matches, and 39-percent of those were for people of color. Only about 20-percent of Congress are people of color in the first place. This actually is not surprising, as other recognition tools have issues with people of color, so seeing that Rekognition is also having the issue, is not a big surprise.

Jacob Snow, one of the technology attorneys for the ACLU stated that "I think these results emphasize that there are applications of face surveillance that are not safe. You could imagine a police officer getting a match that somebody has, for example, a concealed weapons arrest and that's going to bias the police officer." While facial recognition could really help law enforcement catch criminals that are on the loose and identify them. It could also falsely accuse a number of people, especially those of color. Which wouldn't be a good look, and currently the police already have a lot against them in the US. And this is something else that they don't need to worry about right now.

You May Like These
More Like This:
Android Headlines We Are Hiring Apply Now