GULC Report Proposes Police Database Oversight

Across the nation, law enforcement agencies are using facial recognition networks containing the photographs of about half of American adults, posing potential privacy and civil liberty violations, according to an Oct. 18 report released by the Center on Privacy and Technology at the Georgetown University Law Center.

The study calls for greater oversight from Congress and local government, as the current lack of regulation creates a climate in which officers can perform searches on the networks without warrants. Officers can also currently access a database that includes the driver’s license photos of people from 26 states.

The report, entitled “The Perpetual Line-Up: Unregulated Police Face Recognition in America” found that over 117 million people are included in the facial recognition network.

The network is compiled mainly from a database with state IDs, police bookings and real-time scans from cameras in public locations.

The report found that no agency required officers to acquire a warrant before conducting a search of the database and often did not require officers to suspect someone of committing a crime before using the database to identify them.

Law enforcement agencies use facial recognition systems to cross-reference existing images of people entered within the networks. The system produces a virtual lineup of suspects to help police identify suspects in a crime.

The report also claims that racial profiling runs rampant, as database algorithms disproportionately target black individuals.

The Supreme Court has yet to rule on any cases regarding the use of facial recognition in criminal cases, leaving broader questions about the system’s interference with Fourth Amendment rights, which protects citizens from unreasonable search and seizure.

GULC Dean William Treanor said the study, which is the result of a yearlong investigation, is not only a significant discovery, but also reveals the importance of continuing to examine the intersection between privacy, technology and law.

“[The report] represents a major step in how we think about the use of face recognition technology and how it is regulated,” Treanor wrote in an email to The Hoya. “This work once again highlights the critical need to have lawyers who understand technology and are well trained in various aspects of technology law.”

To conduct the report, the researchers requested records from more than 100 law enforcement agencies across the country. Only 52 agencies responded with an acknowledgement of using facial recognition services. Of these 52, only one provided evidence of auditing their officers’ searches on the facial recognition system for misuse.

Alvaro Bedoya, the Center on Privacy and Technology executive director and one of the study’s initiators, said this system gives law enforcement access to a massive lineup in fingerprint or DNA databases.

“You are not going to be in a criminal DNA database,” Bedoya said in an Oct. 23 interview with NPR’s “Weekend Edition.” “Yet by simply standing for a driver’s license photo, 26 states enroll you in basically a virtual lineup just like in the movies, except it’s not a human being pointing to the suspect. It’s an algorithm.”

One of the researchers for the project, Harrison Rudolph (GRD ’16), who is also a GULC Center on Privacy and Technology law fellow, said this was particularly disconcerting as most of the citizens included in the database are there regardless of previous criminal activity.

“Historically, law enforcement biometrics databases, like fingerprint databases, have been primarily made up of criminals,” Rudolph said. “But driver’s license photo databases are primarily composed of law-abiding citizens and I think that is really a startling finding.”

According to American Civil Liberties Union Senior Policy Analyst Jay Stanley, this technology could also potentially infringe on citizens’ privacy and exponentially expand the power of local law enforcement.

“Where you really get into the serious privacy issues is where you have real-time searches against people on the streets, attaching it to video cameras, and that really has a potential to add up to an infrastructure for total surveillance for everybody, all the time,” Stanley said.

GULC found that this inclusion of citizens into facial recognition databases has been happening for at least the past few years and has gone largely unnoticed because the information is predominantly kept private.

Rudolph noted that the lack of a historical precedent, in addition to a lack of Supreme Court rulings, has precluded any states from enacting comprehensive laws regulating facial recognition.

According to Rudolph, this has manifested itself in a lack of transparency from law enforcement, which has prompted an explosion in the use of facial recognition databases, most of which is unknown to the public.

“What we found was that, in short, face recognition is out of control. There are no rules, no audits, no transparency, few accuracy tests and no comprehensive laws, and so it’s difficult to find out about police use of face recognition when they’re not making their use public,” Rudolph said.

The report further cited a 2012 study co-authored by Michigan State University researchers and the FBI, which found facial recognition systems were significantly less accurate when identifying young people, women and especially African Americans.

African Americans are also disproportionately represented in the database because of the racial disparities in arrest rates. According to the National Association for the Advancement of Colored People, African Americans are incarcerated nearly six times the rate of whites.

Rudolph said biases implicit in the system could result in a large volume of citizens being falsely identified as suspects.

“The concern is that facial recognition is less accurate than other biometrics technologies like fingerprints, it makes mistakes, and it doesn’t make mistakes equally on everybody,” Rudolph said. “If you belong to one of those groups, facial recognition technology tends to work less well on you and that means there’s a greater likelihood that you’ll be falsely identified as a suspect in a system that doesn’t give no for an answer.”

The Center on Privacy and Technology has provided proposals to combat these charges, including model legislations that could be passed by both state and federal governments as well as a model policy for police departments to adopt.

“The public should be put on notice about the use of facial recognition technology,” Rudolph said. “Another recommendation is that facial recognition companies should engage in accuracy tests of their algorithms to check for bias, particularly on the basis of race.”

Congressional Oversight and Government Reform Committee Chairman Jason Chaffetz (R-Utah) lauded the report for its proposed solutions and emphasized the importance of Congress acknowledging facial recognition as a tool, but also regulating its use.

“Safeguards must be in place to ensure its accuracy and to identify and eliminate any potential bias or deficiencies,” Chaffetz wrote in an email to The Hoya. “The technology must be used in a manner consistent with our Constitutional right of protection against unwarranted government searches. Continued legislative oversight is needed to ensure proper use of this powerful emerging technology. I applaud the good work that went into preparing this significant report.”

Stanley stressed that as facial recognition databases push law enforcement techniques into uncharted legal territory, the law will need to adapt to new questions about basic civil liberties.

“We have never seen anything like what facial recognition technology can do in all of human history. It’s got the potential to really change our country,” Stanley said. “We need to be very careful about how we allow it to be used.”

Have a reaction to this article? Write a letter to the editor.

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>