The NYPD has used face recognition since at least 2011 but refuses to provide any information about how the technology is used.
Researchers at Georgetown University law school filed a Freedom of Information lawsuit against the New York City Police Department today for the agency’s refusal to disclose documents about its longstanding use of face recognition technology. The NYPD’s face recognition system, which has operated in the department’s Real Time Crime Center since at least 2011, allows officers to identify a suspect by searching against databases of stored facial photos.
Records pertaining to the NYPD’s program were requested in January 2016 by researchers at Georgetown Law’s Center on Privacy and Technology as part of The Perpetual Line-Up, a year-long study on law enforcement uses of facial recognition technology. After receiving public records from more than 90 agencies across the country, the study found that one in every two American adults is enrolled in a criminal face recognition network and that “few agencies have instituted meaningful protections to prevent the misuse of the technology.”
Despite the fact that numerous agencies disclosed similar information about policies, procedures, training, audits, contracts, and agreements relating to their use of facial recognition technology, the NYPD determined in January 2017 that it was unable to find any records responsive to the Center’s detailed records requests.
Instead, the NYPD sent the researchers a single memo outlining how officers should use the results of a facial recognition search, which confirms that the department has a specific unit, staffed with analysts, actively conducting facial recognition searches. The department also acknowledged that it located records relating to the purchase of facial recognition technology, but it denied access to those records in their entirety, according to the lawsuit filed today.
Clare Garvie, one of the co-authors of Georgetown’s report and an expert on face recognition technology, described the NYPD’s lack of transparency as a “very worrying prospect” given the technology’s potential for invasive surveillance, including in real time. Georgetown’s research, among others, has shown that the technology can make mistakes, meaning that innocent people may be investigated and charged for crimes they did not commit. In 2015, a spokesperson for the NYPD said that the technology had “misidentified” five people.
No federal laws currently regulate law enforcement’s use of face recognition systems. Because the NYPD’s own policies, manuals, and documents are “the only controls” on its own system, their disclosure is in the public interest, Garvie explained.
“If no records exist, that means that there are no controls on the use of face recognition technology and we ought to worry about that. If there are records, then why did the Police Department say that it couldn’t find them?” said David Vladeck, a member of Georgetown’s law faculty, in a press release.
Publicly available information has repeatedly indicated not only that the department’s program exists, but that corresponding documentation exists as well. The department’s own personnel have boasted of the program’s specific statistics to media, international experts have published on the department’s practices, LinkedIn profiles explicitly describe employee training protocols, and two of the agency’s vendors have revealed that they supply the NYPD. For instance, a document disclosed to Georgetown researchers by the Los Angeles County Sheriff’s Department revealed that the vendor DataWorks Plus also provides “a fully integrated facial recognition solution for the New York City Police Department … for over 2 million records and 12,000 web users.” All of these descriptions, the lawsuit contends, suggest the presence of records responsive to the Georgetown Center’s request.
“We have no idea whether a certain legal standard is required prior to a face recognition search, or if it is used to investigate certain crimes,” Garvie said. “We similarly have no idea what databases are searched — who is in the NYPD’s perpetual lineup; what level of training the analysts conducting the searches receive; whether there are any quality controls on the images submitted for search.”
The NYPD’s secrecy is especially concerning given that the department is currently in the process of expanding its face recognition program. In October, Governor Andrew Cuomo announced plans to use face recognition technology to identify up to 80,000 drivers traveling in and out of New York City daily: “At structurally sensitive points on bridges and tunnels, advanced cameras and sensors will be installed to read license plates and test emerging facial recognition software and equipment.”
While experts like Garvie concede that the technology can greatly benefit police investigations, they are concerned that New York City’s latest expansion is being rolled out, like the rest of the department’s face recognition system without oversight, accuracy testing, or public debate.
The NYPD did not respond to a request for comment.
Top photo: A security camera is mounted on the side of a building overlooking an intersection in midtown Manhattan, on July 31, 2013.