Law enforcement agency, FBI (Federal Bureau of Investigation) is piloting Amazon’s facial matching software, Amazon Rekognition, as a way to sieve through tons of video surveillance footage the agency regularly accumulates during investigations. As per the FBI officials, the pilot system kicked off in early 2018 following a cord of high-profile counterterrorism investigations that experienced the boundaries of the FBI’s technological capabilities.
In the year 2017, for example, mass shooting in Las Vegas where the law enforcement agency collected a petabyte amount of data, much of it video from cell phones and surveillance cameras. FBI Deputy Assistant Director for Counterterrorism, Christine Halvorsen stated that they had agents and analysts, eight per shift, working 24/7 for three weeks going through the video footage. Halvorsen made those remarks in November at the AWS re: Invent conference in Las Vegas, where she explained how the FBI is utilizing Amazon’s Cloud platforms to do counterterrorism investigations. She further said that Amazon Rekognition could have gone through the same trove of data from the Las Vegas shooting in 24 hours or three weeks faster than it took human FBI agents to detect every instance of Paddock’s face in a lot of videos.
Halvorsen said that noting the technology like Amazon Rekognition frees up the FBI agents and analysts to implement their skills to other facets of the investigation or other cases. As Amazon is now a major technology provider to the government agencies, much of it through its Cloud business, AWS, includes the CIA and Defense Department as customers; it is less clear how its AI software is being utilized in the public sector.