fb-pixel Skip to main content

Surveillance technology used by police gets fresh scrutiny after nationwide protests, looting

Tools to identify faces, scoop up smartphone data raise constitutional issues

A protester used his attire to express opposition to facial-recognition technology during a demonstration in London last month.Kirsty Wigglesworth/Associated Press

In response to the violence that broke out during some George Floyd protests, the FBI has set up a Web page where people can submit videos and photos. The agency hasn’t said, but some advocates suspect it will use facial-recognition software to analyze the images and identify suspects.

But members of the Boston City Council want no part of this.

On Tuesday, a City Council subcommittee took up a proposed ordinance to ban the use of facial-recognition software by the city’s agencies, including the Police Department. The measure, sponsored by Councilors Michelle Wu and Ricardo Arroyo, is expected to come to a vote before the full council in the next few weeks.


“Facial recognition tends to further racial inequity," said Arroyo, because the software frequently misidentifies people with dark skin. A 2019 federal study found that facial-recognition programs consistently made far more mistakes identifying pictures of dark-skinned people than with pictures of light-skinned people.

At the hearing, Police Commissioner William Gross said his department does not use facial recognition because of the shortcomings. “As an African American male," he said, "the technology that is in place today does not meet the standards of the Boston Police Department, nor does it meet my standards.”

Nonetheless, Gross argued against a total ban, saying the technology may improve enough some day to be a vital lifesaving aid. “Any prohibition on the usage of these tools without a full understanding of potential uses . . . could be harmful and impede our ability to protect the public," he said.

Brookline, Cambridge, Northampton, Somerville, and Springfield have already banned local government use of facial recognition.

And foes of the technology won a surprising new ally as the computing giant IBM Corp. announced late Monday that it was shutting down its facial-recognition business. Chief executive Arvind Krishna called for "a national dialogue on whether and how facial-recognition technology should be employed by domestic law enforcement agencies.”


Then on Wednesday, the Internet giant Amazon announced it would halt police use of its Rekognition facial-recognition system for one year, except for use in investigations of human trafficking. Amazon said it hoped the moratorium would give Congress time to draft regulations to govern use of the technology.

Massachusetts State Police use the technology “to identify specific unknown criminal offenders,” according to a spokesman. In addition, there’s no uniform policy on the use of facial-recognition software by the nation’s 18,000 federal, state. and local law enforcement agencies.

The FBI did not respond to return a request for comment.

Privacy activist Joy Buolamwini, founder of the Algorithmic Justice League and a doctoral candidate at the Massachusetts Institute of Technology, said governments don’t even have to tell citizens they may be subject to facial identification.

“So many agencies can use and purchase these kinds of technology without any kind of scrutiny,” Buolamwini said. “There’s no federal law that requires the disclosure of this technology, and so we’re operating in the dark.”

It’s also impossible to know what kind data the police may be gathering from the air. Members of a US House oversight committee are demanding that the Customs and Border Protection agency explain why it flew an automated Predator drone above Minneapolis during that city’s protests. Such aircraft could carry high-resolution spy cameras as well as radio interception gear to track the sources and destinations of wireless phone calls.


And that’s not the only high-end technology that police are bringing to bear.

A 2016 survey of US law enforcement agencies found that 70 percent monitor services such as Facebook and Twitter to assist in criminal investigations. The FBI has arrested one Illinois man and two others from Missouri who allegedly traveled to Minneapolis to participate in rioting there. The FBI said one suspect posted video of himself engaged in looting. Another man posted messages that allegedly sought to incite a riot. A third was charged with distributing information that could be used to make a weapon of mass destruction, after he posted a recipe for making the incendiary chemical napalm.

Police agencies have also created false online identities to infiltrate activist groups.

In 2016, a Memphis police officer began forming online friendships with hundreds of Black Lives Matter activists via Facebook, in order to track their activities. A federal court in 2018 banned the practice, saying it violated a 1978 consent decree forbidding the city’s police from spying on political groups. But the consent decree covers only the Memphis police. In the rest of the country, there is no law against the practice. In fact, a federal court ruled in 2014 that police in New Jersey could use evidence obtained by tricking a criminal into friending a police officer with a fake identity on Instagram.

Law enforcement agencies also work with commercial, for-profit intelligence services that scoop up and analyze social media messages in bulk. The Intercept, an online magazine, reported that the Department of Homeland Security partnered with Virginia-based LookingGlass Cyber Solutions in 2018 to track Facebook messages from people who oppose President Trump’s immigration policies. The information was shared with state law enforcement agencies to help them plan for hundreds of demonstrations throughout the United States. LookingGlass did not respond to requests for comment.


Even a looter’s smartphone could betray him: There’s a good chance, for example, that a photo of a looting scene will be automatically uploaded to a cloud-storage service such as Google Photos or Apple iCloud. Even if he deletes the image from the phone, the police can still subpoena the cloud files.

Or the phone may be subject to a “geofence warrant,” a controversial kind of search warrant that lets police scoop up GPS location data from every phone near a crime scene.

Say a store in Downtown Crossing is looted at 10 p.m. Boston police could ask a judge to issue a geofence warrant for every phone that was near store’s location at the time. The Internet search giant Google complies with this type of warrant; Apple did not respond to queries about whether it complies.

In a background briefing with the Globe, Google said that it hands over location data without providing the identities of the phone owners. The police are able to track the movements of each phone over the alloted time, using this information to choose the most likely perpetrators. For instance, one of the phones may have traveled to several other locations that were also looted, making its owner a prime suspect. Google then provides the police with the name of this phone’s owner.


Google says demand for geofence warrants jumped 500 percent last year. But defense attorneys say they violate the Fourth Amendment requirement that a warrant seek specific evidence from a specific suspect. They say geofence warrants are unconstitutional fishing expeditions. A man accused of armed bank robbery in Virginia is challenging the legitimacy of the warrants in a federal court.

In the meantime, Google may well be swamped with geofence warrants from police hoping to track down rioters. When asked if Boston police would seek such warrants, a spokesman refused to answer, saying he could not discuss investigative methods.

Hiawatha Bray can be reached at hiawatha.bray@globe.com. Follow him on Twitter @GlobeTechLab.