Farcical recognition tech? NYPD ‘used Woody Harrelson pic for beer thief search’
In a report called ‘Garbage In, Garbage Out’, Georgetown University’s Center on Privacy and Technology said the incident is just one example of the police department’s flawed software. The NYPD have also used a photo of a New York Knicks player to search for a man wanted in connection to an assault in Brooklyn, the report claims.
The study says security footage from the beer robbery was too pixelated to used by the NYPD software in search of the actual thief, however a search in the database using his three-time Oscar nominee lookalike turned up several possible matches and eventually led to one arrest.
The NYPD says the facial recognition technology is merely a means of producing leads and no one has been arrested on it alone, however researchers say the stakes are too high in homicide, rape and robbery cases to rely on unreliable, or wrong, inputs.
Here’s the full NYPD statement on the report — doesn’t dispute any of the findings. pic.twitter.com/GMwCAzRFl9— Russell Brandom (@russellbrandom) May 16, 2019
“It is one thing for a company to build a face recognition system designed to help individuals find their celebrity doppelgänger or painting lookalike for entertainment purposes,” wrote Georgetown researcher Clare Garvie.
“It’s quite another to use these techniques to identify criminal suspects, who may be deprived of their liberty and ultimately prosecuted based on the match.”
The study says facial recognition has helped the NYPD solve about 2,900 cases in the five years of using the technology. Garvie is set to testify on May 22 in front of the House Oversight Committee on the civil rights issues associated with police use of facial recognition.
4/ Need an example? NYPD detectives use “celebrity comparison” #facerecognition searches. Photo too blurry? No problem—run a search for the suspect’s celebrity doppelgänger instead. pic.twitter.com/2ALvHa43q4— Clare Garvie (@ClareAngelyn) May 16, 2019
Like this story? Share it with a friend!