Policy

Bipartisan thumbs-down to facial recognition technology

Surveillance sparks comparisons to Orwellian dystopia

A Customs and Border Protection officer scans a traveler entering the United States in February 2018 at Miami International Airport. The use of facial recognition technology by the government violates the First and Fourth amendments, some lawmakers believe. (Joe Raedle/Getty Images file photo)

In 2016, police officers in Baltimore used new technology to scan the faces of protesters who filled the city’s streets following the death in custody of Freddie Gray, a 25-year-old black man. Among those whose most recognizable features may have been documented was Rep. Elijah E. Cummings, the Democratic chairman of the House Oversight and Reform Committee.

Three years later, Cummings is still angry such surveillance was conducted without a warrant or reason to believe that he — or any other protester, for that matter — had done anything illegal. Now he’s putting the full weight of his committee’s jurisdiction behind a push to ban facial recognition technology until Congress can pass comprehensive legislation to govern its use.

The rise of software that can recognize and identify faces has long been the stuff of science fiction, but now it’s widely used by private companies and law enforcement agencies despite a complete lack of standards or privacy protections for individuals. Whether facial recognition technology violates someone’s constitutional rights has yet to be decided in court.

Democrats and Republicans both say that if its use continues unchecked, the technology poses a range of threats to everyday Americans, whether through illegal surveillance or misidentification. Because industry-wide best practices haven’t been established, critics say faulty software can find its way into the hands of those who don’t know how to properly use it, potentially with serious consequences.

“Our faces may well be the final frontier of privacy,” Joy Buolamwini, a research assistant at the Massachusetts Institute of Technology and founder of the Algorithmic Justice League, told the committee at a hearing last week.

From a surveillance standpoint, the technology could be used on anyone exercising their right to peaceably assemble. The Baltimore protests surveilled by police were just an example, said Cummings, although Maryland’s facial recognition policies are considered some of the country’s most aggressive.

“You could be at a rally supporting gun rights or protesting gun violence,” Cummings said at the hearing. “You could be marching for the right to life or a woman’s right to choose. You could be pressing for the repeal of the [Affordable Care Act] or the expansion of health care.”

“In all of these cases,” he continued, “the government could monitor you without your knowledge and enter your face into a database that could be used in virtually unrestricted ways.”

Republicans see 1984

Rep. Jim Jordan, the committee’s top Republican, said at the hearing that he wants a “time-out” on using the technology in order to avoid “a ‘1984,’ George Orwell type of scenario.”

Jordan said he thinks the use of facial recognition technology by the government violates the First and Fourth amendments, and further disproportionately and negatively affects minorities.

“And all three of those things, all of that, happens in a country with 50 million surveillance cameras,” he said.

Andrew G. Ferguson, a law professor at the University of the District of Columbia, said even the Constitution is not enough to protect Americans from the reaches of facial recognition technology. Congress must act, he told the committee.

“The Fourth Amendment will not save us from the privacy threat,” he said. “The Supreme Court is making solid strides in trying to update Fourth Amendment principles in the face of these new technologies, but they’re chasing an accelerating train and will not catch up. Only legislation can respond to the real-time threats of real-time technology.”

Since its development, facial recognition software has expanded rapidly. In addition to its use by the FBI and other federal law enforcement agencies, it has been deployed by local police departments in cities around the country and purchased by health companies and even schools.

“Police can’t secretly fingerprint a crowd of people from across the street. They also can’t walk through that crowd demanding that everyone produce their driver’s licenses,” said Clare Garvie of Georgetown University’s Center on Privacy and Technology. “But they can scan their faces, remotely and in secret, and identify each person thanks to face recognition technology.”

In Maryland, for instance, police have access to 7 million driver’s license photos, 3 million state mug shots, and nearly 25 million mug shots in the FBI’s national database, according to the Georgetown Center. The state’s system has not undergone an audit since it was established in 2011.

Experts are also skeptical about the technology’s ability to work correctly. In general, facial recognition software has been found to lack accuracy, especially when identifying minorities.

Buolamwini said that in the course of her research, facial recognition software had failed to recognize even the face of Oprah Winfrey. She noted the technology also misidentified a female Brown University student as a possible suspect in the terrorist bombings in Sri Lanka in April.

The hearing came as a national discussion over limiting the reach of facial recognition technology is gaining momentum. San Francisco recently became the first major city to vote to ban the use of such software by police, and Massachusetts is weighing a similar idea.

Private sector wary

The private sector appears more reluctant to get on board. Last Wednesday, Amazon shareholders rejected a proposal to ban the sale of Rekognition, the company’s facial recognition program which has been used by law enforcement in Oregon and Florida, to governments, according to Reuters.

Shareholders also voted down a proposal to audit Rekognition for privacy and civil rights concerns.

Cummings plans to hold another hearing on the subject on June 4, when the committee will hear from law enforcement agencies who say the technology has benefits. In Maryland, for example, it was used to catch Jarrod Warren Ramos, who killed five people in a mass shooting at the Capital Gazette newsroom in Annapolis last June.

Following that hearing, Cummings says he will direct staffers to identify “sensible and concrete recommendations” that could be crafted into legislation. His ultimate goal is a moratorium on government use of the technology for as long as it takes Congress to legislate its use. “We’re going to get a bill out,” Cummings said last week. “We’ll do the moratorium and give the industry time to perfect what appears to be a defective product.”

Cummings said he would also direct various subcommittees and staffers to investigate agreements between federal and local authorities to share access to facial databases, as well as the use of federal grant money by state and local government to purchase facial recognition software.

In addition to Jordan’s support, Cummings should gain allies among Senate Republicans and Democrats. GOP Sen. Rand Paul, a frequent critic of creeping government surveillance, will wait to see legislative language but said “it sounds like I would be sympathetic.”

Sen. Brian Schatz, a Hawaii Democrat, has introduced legislation with Sen. Roy Blunt, a Missouri Republican, that would ban the use of facial recognition software by private companies unless they receive a user’s explicit consent. But he’s willing to take on the government, too.

“I’m open to prohibiting everything across the board,” Schatz said. “The reason we settled on private business use was that I thought that was more achievable. If this is achievable, I’m certainly open to it.”