Amazon Rekognition was introduced in the year 2016 as a deep learning-based API that can identify “objects, people, text, scenes, and activities” upon providing a photo or video. One of the most impressive features of the API is its accurate facial analysis and recognition.
The next year, Amazon started selling the API to law enforcement agencies including Orlando Police Department in Florida and Washington County Sherrif’s Office in Oregon through Amazon Web Services.
The enforcement agencies shelled $400 to get the data comprising 305,000 mugshot photos in the API and Amazon charges a meager amount of $6 per month for using it, making it cheaper than Netflix.
According to official documents obtained by American Civil Liberties Union of Northern California, the agencies are using the technology to identify and keep a check on the real-time movements of people thus posing a threat to the privacy of the general public. Furthermore, the AI-based technology is error-prone and often commits mistakes in recognizing ‘dark-skinned’ people.
The documents reveal that Rekognition has been used in Police Body Camera to identify people. Massive applications of such a relatively new and imprecise technology have raised concern amongst lawmakers and privacy advocates.
Submitted by: Arnfried Walbrecht