On June 10, 2020, Amazon announced a one-year moratorium on police use of Amazon Rekognition, its facial recognition service. The brief statement said the company was “implementing a one-year moratorium on police use of Amazon’s facial recognition technology,” and added that it hoped the pause would “give Congress enough time to implement appropriate rules” for the technology’s ethical use.
The move came during the nationwide protests over the police killing of George Floyd, in the same week that IBM told Congress it would exit general-purpose facial recognition and Microsoft said it would not sell face recognition to US police without a governing law. Together the three announcements marked an abrupt corporate retreat from a market the same companies had been building.
Amazon’s Rekognition had been a particular target of criticism. In 2018 the ACLU had run members of Congress against a mugshot database using Rekognition and reported 28 false matches, skewed toward people of color, and researchers had published studies finding higher error rates for darker-skinned and female faces. Amazon disputed those tests but faced sustained pressure from civil-liberties groups, shareholders, and its own employees.
The moratorium carved out exceptions, allowing organizations such as Thorn and the International Center for Missing and Exploited Children to keep using Rekognition to fight human trafficking and find missing children. Amazon later extended the police moratorium indefinitely. The episode showed how quickly reputational and political pressure could halt deployment of a profitable AI product - though it left the underlying technology, and its sale to other customers, intact.