Police in Washington are running suspect sketches through Amazon’s controversial facial recognition software, report reveals
- Police use sketches to find criminals with Amazon’s face-scanning software
- The use has caught the attention of skeptics and civil rights groups
- Critics say it increases the risk of falsely accusing someone of a crime
- Amazon will vote on whether to continue selling the tool to police this month
In a previously undocumented use of facial recognition software, police in Washington state are using Amazon’s ‘Rekognition’ to track down criminals with as little as an artist’s sketch.
According to a report from The Washington Post, police in Washington County are able to compare pictures of suspects harvested from security cameras and eye-witness’ cell phone pictures against databases containing 300,000 mugshots of known criminals.
In just Washington County Police Department alone, the report states more than 1,000 facial scans were logged last year which have helped identify subjects, sometimes leading officers to home arrests.
Amazon’s facial recognition software is being used to process criminal sketches in an unprecedented deployment of the technology in law enforcement. File photo
While law enforcement say the software has been a critical tool in expediting investigations and tracking down otherwise elusive criminals, skeptics say the use of facial recognition opens up a proverbial Pandora’s Box of mass surveillance that could lead to more false identifications.
In some cases, the report notes, police are taking the previously undocumented step of running sketches — artist renderings based on eye-witness reports — through the system in hopes of turning up a suspect.
This in particular has riled concern from experts who say that standards for using the technology have yet to catch up to the reality and breadth of its deployment.
When pictures are turned up by the software, they’re accompanied with a percentage match that indicates just how confident the A.I. is that the input image is indeed the suspect.
Amazon has recommended that law enforcement should only act when Rekognition is 99 percent positive that the input picture is a match with a documented mug shot, but there are currently no official guidelines on whose or what data turned up by the computer can be used in an investigation.
Instead, officers are presented with five matches on every search, regardless of how confident the computer is in its ability to identify the suspect.
Despite a growing chorus of concerned researchers, civil rights organizations, and even companies like Microsoft — one of the biggest purveyors of advanced facial recognition software — the technology has continued to find footholds in a law enforcement capacity across the U.S.
By analyzing points on a suspect’s face, Rekognition is able to turn up potential matches in a database containing thousands of mug shots. Stock image
According to a study Georgetown University’s Center for Privacy and Technology in 2016 which surveyed nearly 4,000 state and local law enforcement agencies, 25 percent of respondents reported currently having the ability to be able to run or request facial scans that compare against a database.
Throughout the last several years, adoption rates have likely increased, and with Amazon’s foray into the field, access could get easier — if it’s not blockaded by law or the company’s executive board first.
On May 22 Amazon’s board is set to take a vote on whether to continue selling facial recognition software to governments while several legislatures across the county are in varying stages of potentially passing moratoriums on the technology’s use by public agencies.
Among the most well-documented side effects of facial recognition’s use by law enforcement is the potential to disproportionately affect minorities.
The software not only has a difficult time reading faces with darker complexions, but has been trained mostly on white males, making it more apt to make mistakes on other demographics.
HOW DOES FACIAL RECOGNITION TECHNOLOGY WORK?
Facial recognition software works by matching real time images to a previous photograph of a person.
Each face has approximately 80 unique nodal points across the eyes, nose, cheeky and mouth which distinguish one person from another.
A digital video camera measures the distance between various points on the human face, such as the width of the nose, depth of the eye sockets, distance between the eyes and shape of the jawline.
A different smart surveillance system (pictured) can scan 2 billion faces within seconds has been revealed in China. The system connects to millions of CCTV cameras and uses artificial intelligence to pick out targets. The military is working on applying a similar version of this with AI to track people across the country
This produces a unique numerical code that can then be linked with a matching code gleaned from a previous photograph.
A facial recognition system used by officials in China connects to millions of CCTV cameras and uses artificial intelligence to pick out targets.
Experts believe that facial recognition technology will soon overtake fingerprint technology as the most effective way to identify people.