Opinion | We Built a (Legal) Facial Recognition Machine for $60

Most people pass through some type of public space in their daily routine — sidewalks, roads, train stations. Thousands walk through Bryant Park every day. But we generally think that a detailed log of our location, and a list of the people we’re with, is private. Facial recognition, applied to the web of cameras that already exists in most cities, is a threat to that privacy.

To demonstrate how easy it is to track people without their knowledge, we collected public images of people who worked near Bryant Park (available on their employers’ websites, for the most part) and ran one day of footage through Amazon’s commercial facial recognition service. Our system detected 2,750 faces from a nine-hour period (not necessarily unique people, since a person could be captured in multiple frames). It returned several possible identifications, including one frame matched to a head shot of Richard Madonna, a professor at the SUNY College of Optometry, with an 89 percent similarity score. The total cost: about $60.

“My first reaction was, ‘Oh my god, that is unbelievable,’” Dr. Madonna said, after we reached him and explained the experiment. “I was shocked at how readily it seems that it picked me up, because, really — it’s the side of my head.”

In our exercise, we built a database using only photos from public websites, and we obtained Dr. Madonna’s consent before publishing this story. We’ve deleted the images and data that we collected and are no longer monitoring the Bryant Park cameras.

Over decades, businesses and individuals have installed millions of cameras like the ones we used, inadvertently setting up the infrastructure for mass surveillance. In the past, a human would have to watch the video feed to identify people, making it impossible to comprehensively record everyone’s movements. But the accuracy and speed of modern facial recognition technology means that building a dragnet surveillance system is now feasible.

The law has not caught up. In the United States, the use of facial recognition is almost wholly unregulated.

“The technology has advanced faster than even I thought that it would,” said Jennifer Lynch, surveillance litigation director at the Electronic Frontier Foundation. She said that because of how quickly the technology has advanced, she would now support a wholesale ban on government use of facial recognition.

The cameras in Bryant Park were installed more than a decade ago so that people could see whether the lawn was open for sunbathing, for example, or check how busy the ice skating rink was in the winter. They are not intended to be a security device, according to the corporation that runs the park.

But our experiment shows that a person equipped with just a few cameras and facial recognition technology can learn people’s daily habits: when they arrive at the office each day, who they get coffee with, whether they left work early. When we identified Dr. Madonna, he was on his way to lunch with a job candidate — an example of how the midday outings of even law-abiding citizens can sometimes be sensitive information.

The police and governments may also have access to a vast network of cameras. Combine that with a comprehensive database of faces — like a driver’s license database — and it’s possible to track citizens throughout an entire region in real time. There is no evidence that this is happening on a wide scale in the United States. But that’s not because the technology doesn’t exist. Last year, companies claimed they could compare live feeds to a database of billions of faces.

Authorities have used facial recognition to track down criminal suspects and find missing children. But civil liberties advocates warn about the chilling effect on free speech if the government could monitor everyone’s whereabouts — or, say, identify individuals at a protest. This is not a purely hypothetical concern: During 2016 protests after Freddie Gray died in the custody of Baltimore police, law enforcement used facial recognition on social media images to identify protesters with outstanding warrants.

“Once the government has the ability to track us and identify us wherever we go, it is impossible to speak and participate in society anonymously,” Ms. Lynch said.

New York City is nowhere near China, where the government has installed approximately one surveillance camera for every seven citizens. But according to the A.C.L.U., police here have access to more than 9,000 camera feeds in Lower Manhattan alone.

The M.T.A. has tried using facial recognition on feeds from license-plate cameras at the city’s entry points to identify drivers through their windshields, although those efforts have been unsuccessful so far, according to The Wall Street Journal. And the Department of Transportation already has hundreds of cameras across New York City used to monitor traffic, feeds that are also streamed publicly online.

The traffic cameras are most likely too low-resolution for effective facial recognition. But the city’s LinkNYC kiosks, which are scattered through the streets and intended to provide free wireless internet, each have two security cameras. Law enforcement agencies need a subpoena or court order to gain access to the footage, and using facial recognition is against the policy of the company that owns the kiosks. However, the existence of more than 3,000 additional cameras has raised concerns about their potential to bolster the city’s surveillance capabilities.

Details are sparse, but there is evidence that those capabilities are formidable. The Police Department claims its Domain Awareness System, developed jointly with Microsoft (which also offers facial recognition software), “utilizes the largest network of cameras, license plate readers, and radiological sensors in the world.”

It’s unclear whether the Domain Awareness System currently uses facial recognition, though the Police Department experimented with it in 2012, according to Clare Garvie, an associate at the Center on Privacy and Technology at Georgetown Law School. The police have been reluctant to divulge details, and the center has sued the department for more information.

“We compare facial images picked up by cameras at crime scenes to mugshots in law enforcement records,” said Sgt. Jessica McRorie, a spokeswoman for the department, in an emailed statement. “We do not engage in mass or random collection of facial records from N.Y.P.D. camera systems, the internet, or social media.”

Amazon is one of several companies, including Google and Microsoft, that sell facial recognition services to the public. The company has highlighted positive applications of the service we used, Rekognition, such as its ability to help find lost children. It insists that it requires customers comply with the law and respect others’ rights, but has been criticized for pushing its technology to law enforcement agencies.

Rekognition is already actively used by the sheriff’s office in Washington County, Ore., including to investigate minor crimes like shoplifting. The Orlando, Fla., Police Department is also using the technology in a pilot program.

Amazon notes that its service makes predictions, not decisions, and that the confidence level the service provides should be incorporated in a human review process. The company recommends using a threshold of at least 99 percent for applications of its facial recognition service that involve identification or public safety, though critics of the technology say that the scoring is opaque and that the company has no way of enforcing that threshold. None of the matches we obtained from the Bryant Park footage, correct or incorrect, met the threshold.

Matt Wood, the general manager of artificial intelligence for Amazon Web Services, noted that it is possible that Rekognition, like other types of information available to law enforcement officials, could be used inappropriately. “The law enforcement agency will have to be accountable to these individuals and to the law if they violate people’s civil liberties,” he said. He added that the company has not received any reports of misuse by law enforcement.

In January, however, the A.C.L.U. sent a letter to Amazon asking it to stop selling facial recognition technology to police and government agencies, saying that the company’s attention to civil liberties has lagged behind that of Google and Microsoft.

“Rekognition marketing materials read like a user manual for authoritarian surveillance,” said Nicole Ozer, the technology and civil liberties director for the A.C.L.U. of California, in a statement last year.

In the United States, there are no federal laws that restrict the use of facial recognition. Most states don’t have regulations, nor does New York City, though a city councilman proposed legislation last year that would require businesses to disclose their use of the technology. That would apply to our exercise, but would not extend to law enforcement’s use of facial recognition.

“It’s kind of like a wild, wild west out there,” Ms. Lynch, the E.F.F. lawyer, said.

The lack of regulation has opened the door to a wide range of applications. In 2007, an Arizona sheriff’s office enrolled all of Honduras’s driver’s licenses and mugshots into its database, and a Florida sheriff’s office runs 8,000 searches each month without requiring its officers to have reasonable suspicion of a crime, according to the Georgetown report.

The Georgetown center along with the E.F.F. and others have proposed regulations, including requiring that authorities have reasonable suspicion before conducting a search; prohibiting, except in life-or-death situations, live facial recognition searches using driver’s license databases; and forbidding tracking individuals based on political beliefs, race or religion.

Amazon itself has called for a legal framework that incorporates human review and transparency. But some say that the technology is so dangerous that no regulation is sufficient.

“The future of human flourishing depends upon facial recognition technology being banned,” wrote Woodrow Hartzog, a professor of law and computer science at Northwestern, and Evan Selinger, a professor of philosophy at R.I.T., last year. “Otherwise, people won’t know what it’s like to be in public without being automatically identified, profiled, and potentially exploited.”

Facial recognition is categorically different from other forms of surveillance, Mr. Hartzog said, and uniquely dangerous. Faces are hard to hide and can be observed from far away, unlike a fingerprint. Name and face databases of law-abiding citizens, like driver’s license records, already exist. And for the most part, facial recognition surveillance can be set up using cameras already on the streets.

It might be too late for a moratorium or ban, however. Facial recognition is already being used by police departments around the country, Ms. Garvie said.

“We can’t lock law enforcement agencies into 20th-century technology just because 21st-century technology raises very serious risks,” she said.

Dr. Madonna, the person we identified, said he understood that tension. He was initially astonished when we reached out to him, but he said that as a doctor, he often talks to students about the ratio of risk to benefit. He saw the tremendous benefits that facial recognition could offer, he said.

But the technology is open to abuse, he added, when individuals or governments can use facial recognition to track any group, or just about any ordinary citizen — even someone walking through Bryant Park.

Designed by Jessia Ma. Video animations by Drew Jordan.

Sahil Chinoy is a graphics editor for The New York Times Opinion Section.

The Times is committed to publishing a diversity of letters to the editor. We’d like to hear what you think about this or any of our articles. Here are some tips. And here’s our email: letters@nytimes.com.

Follow @privacyproject on Twitter and The New York Times Opinion Section on Facebook and Instagram.