Face recognition in law enforcement
Face recognition in law enforcement is a method of cataloging and identifying human faces in order to track people suspected of illegal activity and make arrests. Police often collect mugshots or driver’s license photos to build local, state, and federal face recognition databases. Law enforcement can query these databases to identify people from mugshots, CCTV, or traffic cameras. The software used to return matching database results analyzes patterns based on a person’s facial features and geometry. The facial recognition and biometrics market for federal, state, and local law enforcement in many countries was a $136.9 million dollar industry in 2018, and is expected to increase to $375 million by 2025. These technologies have been developed for law enforcement since the 1990s by multiple private companies and government agencies, and they are particularly controversial for both their likelihood of misidentifying individuals and the potential for the technology to create a surveillance state. The use of face recognition technology for law enforcement has been banned in the American cities San Francisco, Oakland, Berkeley, and Somerville, and there are proposed bans in many other parts of the world.
- 1 History of Development
- 2 Face Recognition Technologies and Companies
- 3 Law Enforcement Applications
- 4 Ethical Concerns
- 5 References
History of Development
In 1993, the American Defense Advanced Research Projects Agency (DARPA) and National Institute of Standards and Technology (NIST) introduced the Face Recognition Technology (FERET) program to develop automatic face recognition capabilities that could be employed to assist security, intelligence, and law enforcement personnel in the performance of their duties. The project involved creating a database of face images and developing face recognition algorithms. Building off of FERET in the early 2000s, the NIST began Face Recognition Vendor Tests (FRVT), which were designed to provide independent government evaluations of commercial and prototype facial recognition systems. These evaluations were designed to provide law enforcement agencies in the United States with information to determine the best ways to deploy facial recognition technology.
In 2001, the Pinellas County Sheriff's Office in Florida created a forensic database that allowed police officers to search into the photo archives of Florida’s Department of Highway Safety and Motor Vehicles (DHSMV). By 2011, around 170 officers had been provided with cameras that allowed them to take pictures of suspects that could be cross-checked against the database.
In 2011, the government of Panama, partnering with the former U.S. Secretary of Homeland Security Janet Napolitano, authorized a pilot program of FaceFirst’s facial recognition platform in order to cut down on illegal activity in Panama’s Tocumen airport. The system resulted in the apprehension of multiple Interpol suspects and has since expanded. The use of FaceFirst in the Tocumen airport is one of the first and largest installations of face recognition in an airport.
Face Recognition Technologies and Companies
Clearview AI is an American technology company that provides face recognition technology primarily to law enforcement agencies. The company has created a database of more than 3 billion images scraped from the internet, including social networks like Facebook, Instagram, and YouTube. Primary clients of Clearview AI include Immigration and Customs Enforcement (ICE) and the U.S. Attorney’s Office for the Southern District of New York. Individuals and employees can also have access to Clearview’s databases, regardless of their organization's formal relationship with the company. Despite the NYPD having no institutional relationship or contract with Clearview, internal logs show as many as 30 officers within the department conducted 11,000 searches using the software. While Clearview AI is marketed primarily to law enforcement agencies, they also provide their software to companies like Macy’s, Walmart, Eventbrite, the NBA, Coinbase, and Equinox. In 2020, multiple social media companies, including Twitter, YouTube, Facebook, and LinkedIn, sent cease and desist letters to Clearview AI and demanded the company stop using data scraped from their social networks for its database. 
Amazon Rekognition is a cloud-based computer vision platform that was created by Amazon's Web Services (AWS) division and launched in 2016. It is intended to provide “highly accurate facial analysis and facial search capabilities that you can use to detect, analyze, and compare faces for a wide variety of user verification, people counting, and public safety use cases.” The software has been sold and used for numerous United States government agencies, including ICE and the Department of Homeland Security, and multiple private companies. After running a pilot program of the Rekognition software for over a year, the Orlando police department ended its use of the technology over technical errors and ethical concerns. In 2018, Amazon employees reportedly sent a letter to Amazon CEO Jeff Bezos requesting that Amazon stop selling facial recognition software to US law enforcement.
Next Generation Identification (NGI)
Next Generation Identification (NGI) is a Federal Bureau of Investigation (FBI) project intended to expand the FBI’s Integrated Automated Fingerprint Identification System (IAFIS) criminal and civil fingerprint database to include multimodal biometric identifiers such as iris scans, palm prints, face recognition-ready photos, and voice data and make that data available to other agencies at the state and federal levels. NGI’s facial recognition search offers an automated search and response system targeted toward state and local law enforcement, where authorized law enforcement may submit a photo for a search against over 30 million criminal mug shot photos and receive a list of ranked candidates as potential investigative leads. The system was designed and built by Lockheed Martin Corp. under a contract awarded in 2008.
Megvii Technology is a Beijing-based artificial intelligence startup known for its facial recognition software Face++, which is the world’s largest open-source computer vision platform. Megvii is valued at over $4 billion as of 2019, and its main clients include Alibaba, Ant Financial, Lenovo, China Mobile, and multiple Chinese government entities.
Law Enforcement Applications
China is considered to be the world leader in face recognition technology. As of 2018, 170 million CCTV cameras are already in place, and an estimated 400 million are planned to be built by 2021. The government has also expanded their program to include a facial recognition sunglasses program for police on the outskirts of Beijing. China’s use of this technology has been especially criticized for its role in the surveillance and detainment of ethnic minorities, specifically the Muslim Uighur minority in Xinjiang.
In early 2020, the European Union was expected to enact a temporary five-year ban on the use of facial recognition technology in public places so that the technology's impact could be studied before implementation. However, the final draft, completed in February 2020, no longer includes a ban due to fears that it would stifle innovation and compromise national security. The legislation instead outlines a definition for “high-risk” AI applications that can interfere with citizen’s rights, such as those used in the fields of transportation, healthcare, and law enforcement. Without the ban, individual member nations will be responsible for regulating facial recognition.
In January 2020, Germany’s Interior Minister announced plans to implement facial recognition systems at 134 railway stations and 14 airports. The plan has not yet been officially confirmed and has been the subject of criticism by activist groups like the “Face Recognition Stop” alliance over the risk of misidentifications.
India currently has plans to create one of the world’s largest facial recognition systems, aiming to build a single, centralized database that could be accessed across all 29 states and seven union territories. The system would match faces from India’s network of CCTV cameras against an image database consisting of mugshots, passport photos, government-issued identity cards, driving licenses, and images from other agencies. The facial recognition system currently used by the government in New Delhi was initially acquired to identify missing children. In March 2020, law enforcement agencies in India used facial recognition to identify more than 1,100 people who took part in the New Delhi riots that occured on February 25 and 26 of 2020.
Face recognition trials began in Moscow in 2017, using technology from the Russian firm NtechLab to scan footage from Moscow’s network of 160,000 CCTV cameras. NtechLab is best known for its FindFace software, which was launched in 2016 and allows users to match anyone in a picture to their profile on VKontakte, a Russian social media platform. The consumer app was shut down in 2018 after it was used to dox and harass sex workers, and NtechLab transitioned the technology’s use to primarily enterprise and government work. In May 2019, Moscow announced plans to install facial recognition software in up to 200,000 surveillance cameras around the city, with 105,000 connected by the end of 2019.
In January 2020, lawyer and activist Alena Popova and opposition politician Vladimir Milov of Russia’s Solidarnost party filed a case against Moscow’s Department of Technology (DIT), which manages Moscow's video surveillance program, seeking to ban the use of face recognition technology at mass events and protests. The system remained operational during court proceedings to enforce the coronavirus quarantine. In March 2020, a Moscow court ruled that the city’s facial recognition system does not violate the privacy of its citizens.
Face recognition technology is being pursued by multiple local, state, and federal law enforcement agencies in the United States. The first system to take place in the United States began as a project in Pinellas County, Florida in 2001. As of 2016, it is estimated that at least one in four American police agencies could run facial recognition searches. The FBI hosts a Next Generation Identification database with more than 30 million civil and criminal mugshot photos, which can be accessed by many state and local agencies. The FBI also has access to the State Department’s Visa and Passport databases, the Defense Department’s biometric database, and the driver's license databases of at least 16 states, totaling about 412 million images. On a local level, the city of Detroit signed a $1 million dollar deal in 2017 for software that will continuously monitor "hundreds of private and public cameras set up around the city." In September 2019, new provisions were added to ensure the police department could not use facial recognition software on live or recorded video or be used to assess a person's immigration status. In San Diego, the TACIDS (Tactical Identification System) program allows law enforcement officers to stop people on the street and use their phones or tablets to take photographs of them and run the images against the county’s mugshot database.
The use of face recognition technology for law enforcement has been banned in San Francisco, Oakland, Berkeley, and Somerville. Oregon and New Hampshire have banned facial recognition on police body cameras, and Maine and Vermont prohibit the use of face recognition with police drones. Maine, Missouri, New Hampshire, Vermont, Washington, Oregon, and Hawaii all have restrictions on law enforcement’s ability to use driver’s license databases for facial recognition systems.
A major concern for use of face recognition technology is that it is prone to error, which can implicate people for crimes they have not committed. The FBI wrote in its privacy impact assessment that its system “may not be sufficiently reliable to accurately locate other photos of the same identity, resulting in an increased percentage of misidentifications.” Even when certain systems can accurately identify faces that exist in the database, if a subject is not yet in the database it is likely a system will produce false positive matches.
Multiple studies have shown that many facial recognition systems have a much higher rate of error for people of color, especially women of color. In 2017, two MIT researchers, Joy Buolamwini and Timnit Gebru, published a study that found that systems developed by Microsoft, IBM, and the Chinese firm Megvii had an error rate for gender recognition for women of color ranging from 23.8% to 36%, whereas lighter-skinned men had an error rate between 0.0 and 1.6%. Another study found similar problems with Amazon’s face recognition technology Rekognition, which has been used by multiple US agencies. It has also been criticized for being used disproportionately against people of color. In 2016, the ACLU revealed that US police agencies had been monitoring protestors and activists by running social media photos through third-party facial recognition software. The system was used during the protests in Ferguson, Missouri following the death of Michael Brown and the protests in Baltimore following the death of Freddie Gray.
Face recognition technology has also been criticized as being a breach of privacy and escalating the possibility of enabling a surveillance state. Some claim that face recognition inherently undermines the freedom of citizens by enabling complete surveillance of everyone, all the time.
In 2018, a controversy in India began when opposition politician Ajay Maken accused the New Delhi government of having awarded a contract to provide about half of the CCTV cameras it plans to install in the capital to Prama Hikvision, a joint venture between the Chinese company Hikvision and the Indian company Prama Technologies, which would put citizens at risk for espionage. Ashish P. Dhakan, Prama Hikvision's CEO, confirmed that the company was awarded the contract but denied any unauthorized collection of information or espionage. Hikvision, however, has been barred from selling technology to U.S. federal government agencies and been connected to the detainment of roughly a million people without formal charges, mostly from the Muslim Uighur minority in Xinjiang.
- Georgetown Law Center on Privacy & Technology. "The Perpetual Line-up: Unregulated Police Face Recognition in America" Retrieved on 5 March 2020.
- Steve Symanovich. "How does facial recognition work?" Retrieved on 5 March 2020.
- Andrew Heinzman. "How Does Facial Recognition Work?" Retrieved on 5 March 2020.
- Claire Merchlinsky. "How facial recognition became a routine policing tool in America" Retrieved on 5 March 2020.
- NIST. "Face Recognition Vendor Test (FRVT)" Retrieved on 25 March 2020.
- Sean Captain. "Portland plans to propose the strictest facial recognition ban in the country" Retrieved on 5 March 2020.
- Jesse Davis West. "A Brief History of Face Recognition" Retrieved on 25 March 2020.
- Taylor Hatmaker. "Clearview said its facial recognition app was only for law enforcement as it courted private companies" Retrieved on 25 March 2020.
- Jon Porter. "Facebook and LinkedIn are latest to demand Clearview stop scraping images for facial recognition tech" Retrieved on 25 March 2020.
- AWS. "Amazon Rekognition" Retrieved on 25 March 2020.
- Ben Gilbert. "Amazon sells facial recognition software to police all over the US, but has no idea how many departments are using it" Retrieved on 25 March 2020.
- Abrar Al-Heeti. "Orlando police are ditching Amazon Rekognition" Retrieved on 25 March 2020.
- Kate Conger. "Amazon Workers Demand Jeff Bezos Cancel Face Recognition Contracts With Law Enforcement" Retrieved on 25 March 2020.
- EFF. "FBI's Next Generation Identification Biometrics Database" Retrieved on 25 March 2020.
- FBI. "Next Generation Identification (NGI)" Retrieved on 25 March 2020.
- Alice Lipowicz. "FBI deploys faster fingerprint ID system" Retrieved on 25 March 2020.
- Eamon Barrett. "In China, Facial Recognition Tech Is Watching You" Retrieved on 25 March 2020.
- Catherine Shu. "Megvii, the Chinese startup unicorn known for facial recognition tech, files to go public in Hong Kong" Retrieved on 25 March 2020.
- Shannon Liao. "Chinese police are expanding facial recognition sunglasses program" Retrieved on 5 March 2020.
- Cate Cadell. "Hikvision, a surveillance powerhouse, walks U.S.-China tightrope" Retrieved on 5 March 2020.
- Amrita Khalid. "The EU’s agenda to regulate AI does little to rein in facial recognition" Retrieved on 25 March 2020.
- Philipp Grüll. "Germany’s plans for automatic facial recognition meet fierce criticism" Retrieved on 5 March 2020.
- Julie Zaugg. "India is trying to build the world's biggest facial recognition system" Retrieved on 5 March 2020.
- Manish Singh. "https://techcrunch.com/2020/03/11/india-used-facial-recognition-tech-to-identify-1100-individuals-at-a-recent-riot/" Retrieved on 5 March 2020.
- James Vincent. "Moscow rolls out live facial recognition system with an app to alert police" Retrieved on 25 March 2020.
- Anthony Cuthbertson. "Porn Stars and Sex Workers Targeted With Facial Recognition App" Retrieved on 25 March 2020.
- Felix Light. "Russia Is Building One of the World’s Largest Facial Recognition Networks" Retrieved on 25 March 2020.
- BBC. "Russia's use of facial recognition challenged in court" Retrieved on 25 March 2020.
- Alexander Marrow. "Russian court rules in favor of facial recognition over privacy claims" Retrieved on 25 March 2020.
- Jennifer Lynch. "New Report: FBI Can Access Hundreds of Millions of Face Recognition Photos" Retrieved on 5 March 2020.
- Joe Guillen. "Detroit police oversight board approves controversial facial recognition policy" Retrieved on 5 March 2020.
- Sasha Ingber. "Major Police Body Camera Manufacturer Rejects Facial Recognition Software" Retrieved on 5 March 2020.
- Ali Winston. "Facial recognition, once a battlefield tool, lands in San Diego County" Retrieved on 5 March 2020.
- Jon Schuppe. "Facial recognition gives police a powerful new tracking tool. It's also raising alarms." Retrieved on 5 March 2020.
- Rand Corporation. "Biometrics Expert Delivers Lecture on Facial Recognition" Retrieved on 5 March 2020.
- EFF. "Face Recognition" Retrieved on 5 March 2020.
- [Joy Buolamwini, Timnit Gebru. "Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification" Retrieved on 5 March 2020.
- Terry Collins. "Facial recognition: Do you really control how your face is being used?" Retrieved on 5 March 2020.
- Alex Hern. "What is facial recognition - and how do police use it?" Retrieved on 5 March 2020.