Issue: Facial Recognition

The New York City Police Department (NYPD) arrested a 31-year old man after he tried to rape and assault a woman on the subway at a station in midtown Manhattan. The NYPD facial recognition team utilized a video taken by a bystander to identify the man, based on a photo taken from a previous arrest.

A male victim was shot and seriously injured during a robbery in Detroit, Michigan. Using images from high-quality video surveillance covering the location, investigators aided by facial recognition software were able to identify the suspect, who was later taken into custody after further investigation.

A 15-year old girl in Scranton, Pennsylvania, was sexually assaulted by an adult male she met online. Beyond seeing him in person, the only additional information she had was from his online profile. Police were able to use facial recognition on one of the digital images to provide some potential matches from a state database, from which the victim was able to identify a likely match. After additional investigative work, authorities obtained a search warrant for the home of the identified suspect, who later admitted to the crime.

In Detroit, Michigan, in 2019, three members of the LGBTQ+ community were shot and killed by a man at a local home. The Detroit Police Department used facial recognition – in combination with other investigative tools – to help identify the suspect based on video images from a nearby gas station. The suspect was charged with three counts of murder, in addition to other charges.

Facial recognition contributes to more accurate identification. The National Institute of Science and Technology has found that forensic examiners performed best when supported by facial recognition technology and the most accurate performance resulted when these efforts are combined. Proper use of this technology is critical to protecting the innocent and limiting effects from inherent human biases, as eyewitness identifications in criminal investigations are notoriously prone to error. According to the Innocence Project, mistaken eyewitness identifications have been the key factor in 71 percent of wrongful convictions in the U.S. later overturned.

During and post-COVID-19, there is considerable opportunity to use facial matching technology for access control to help ensure security and safety. Many hospitals have already implemented facial recognition to eliminate the need for front-line health care workers to swipe badges or type in codes to verify their credentials. Reducing contact with potentially contaminated surfaces is key when staff must be on high alert around the spread of germs and disease. Such technologies also can perform dual security and hygienic functions in more common settings as well, providing touchless access to virtually any workspace and giving an added layer of hygienic security, as opposed to handing over a pass card or touching a fingerprint scanner.

Facial recognition technology is deployed in dozens of airports across the United States and continues to grow. U.S. Customs and Border Protection, along with airport officials, match passport photos to a database to verify the identity of thousands of travelers entering and leaving the U.S. each week. The technology is proving to be an important tool for border security. As of June 2020, nearly 300 individuals have been intercepted attempting to enter the U.S. under a fraudulent identity. Beyond violating laws against illegal entry, passport fraud fuels many other types of crime, including human trafficking and drug smuggling.

In 2018, detectives in Munster, Indiana, tried to identify a suspect who had attempted to rob a local business at gunpoint, after releasing a photo from the location’s surveillance system which was shared by local media. No leads were generated until facial recognition was used to find a possible match, a man who had skipped parole after serving a prison sentence for nine armed robberies in Illinois. The suspect was identified after the store owner was shown a photo lineup that included his picture, and the suspect was arrested several months later. Without facial recognition, the suspect would likely never have been found.

Jarrod Ramos was angered by a story the Capital Gazette (a newspaper headquartered in Annapolis, Maryland) ran about him in 2011 and brought a lawsuit against the paper for defamation, which a judge later dismissed. In 2018, Ramos entered the Gazette's office building with a shotgun and killed five employees, leaving two others critically injured. Anne Arundel County Police obtained an image of Ramos and sent it to the Maryland Combined Analysis Center, which helped identify him by comparing the photo to others in the Maryland Image Repository System.

The Kansas Department of Revenue’s use of facial recognition software in 2014 led to the investigation of the largest forced labor trafficking case in the United States, all through identifying cases of driver’s license fraud in their database.

For over a decade, federal, state and local law enforcement have effectively used facial recognition technology in thousands of investigations. Many public safety officials feel that this technology is becoming a game-changer for keeping our communities safe, pointing to instances in which crimes would have never been solved or prevented without it. Facial recognition allows investigators to narrow searches for suspects more quickly, find missing children, rescue human trafficking victims, exonerate the innocent, identify the deceased and benefit our communities in many other ways.

In 2017, a man accused of sexually assaulting a minor was apprehended in Oregon after a 16-year manhunt. Using facial recognition technology, the Federal Bureau of Investigation (FBI) was able to identify the suspect after a positive match was found when the suspect sought to acquire a U.S. passport. Similarly, in 2014, the FBI used facial recognition technology to help locate and apprehend a convicted pedophile who had been on the run for 14 years, returning him to New Mexico to face justice.

Since 2015 the nonprofit group Thorn has provided a tool called Spotlight, which uses facial recognition among other technologies to help investigators find underage sex trafficking victims in online ads. Spotlight has reportedly been used in 40,000 cases in North America, helping rescue 15,000 children and identify 17,000 traffickers. For example, after seeing an online post about a missing child from the National Center for Missing and Exploited Children, a law enforcement officer used Spotlight to return a list of sex ads featuring the girl. According to a WIRED story, the girl had been "sold for weeks," and the officer's actions initiated a process that "recovered and removed [her] from trauma."

In New York City in 2019, a man followed a young woman home from work, dragged her into a grassy area and attempted to kidnap and rape her at knife point before eventually letting her go. Investigators used facial recognition technology to compare images from surveillance video at a nearby food store with a mugshot database. Along with additional investigative work, this allowed them to identify a suspect and make an arrest within 24 hours. The 27-year old suspect had previously been arrested for allegedly raping a 73-year old woman but was out on bail.

Facial recognition technology can be used in emergencies in which a suspected terrorist attack is imminent or underway by dramatically reducing the time first responders and investigators need to generate leads and identify suspects. This can improve response times or ultimately prevent such attacks or mitigate further harms, saving lives in the process. For example, as reported by the New York Post, New York Police Department detectives used facial recognition technology to identify a man who sparked terror by leaving a pair of rice cookers in the Fulton Street subway station. Within minutes, detectives pulled still images of a suspect from security footage and used facial recognition software to compare them to mug shots in the NYPD’s arrest database. The system returned several hundred potential matches, and after multiple stages of human review, it took NYPD only one hour to identify the suspect.

In a 40-day span in 2018, and less than 3 weeks after the installation of facial recognition technology in just one location, three impostors were caught at Washington Dulles International Airport using fraudulent passports. As U.S. Customs and Border Protection (CBP) reports, “posing as someone else when attempting to enter the United States is a serious violation of U.S. immigration law.” A 26-year-old man on a flight from Brazil entered Dulles  Airport and presented agents with a French passport. Agents used facial recognition to compare his passport photo to a database of known images with identities, which alerted CBP that the man was carrying a fraudulent passport. The agents discovered the man’s real identification card in his shoe, and it was revealed he hailed from the Republic of Congo, not France. Additionally, a Cameroonian woman posed as an American when trying to enter the United States with a fraudulent passport. The woman presented the CBP officer with U.S. passport in the name of a U.S. citizen, but when the officer ran the passport through the facial recognition technology, it was discovered that their identities did not match. After an additional verification, her true identity was exposed. Similarly, a week prior, a woman was arrested after arriving on a flight from Ghana for posing as a U.S. citizen with a fraudulent passport.

In 2020, police investigated the murder of a man in an Airbnb he had rented with his family. The victim was shot to death over $300 and some computer equipment owed to the suspect. Witnesses who fled the scene produced a Facebook profile of the suspect, allowing officials to verify the identity of the suspect using facial recognition as a secondary tool. Las Vegas Metropolitan Police spokesman Aden Ocampo Gomez said “the technology is strictly a ‘secondary tool’ for investigators that has been in place for a couple of years. The database that police use relies on images of people previously arrested in Metro’s jurisdiction.”