Washington Cops Are Taking a Cue From ‘Fight Club’ for a Secret Facial Recognition Group
Over the last decade, large police forces in Washington state like the Seattle Police Department and the Pierce County Sheriff’s Department have turned to facial recognition technology to identify and track down suspects. But it was, until now, unclear just how widespread the use of the technology was across the state, and whether smaller law enforcement departments had access to those same tools.
Now, thanks to thousands of pages of previously undisclosed emails, OneZero can confirm the existence of a massive, secretive network of police departments working together to share these controversial facial recognition tools. The emails, which date back to at least 2016, also indicate that these departments explicitly tried to keep this cross-department partnership secret from the public. These emails were shared with OneZero from a source who obtained the documents through an open records request.
Many of these cross-department requests in Washington state were made through a previously undisclosed email listserv known as FITlist. FITlist — with FIT standing for “Fraud and Identity Theft” — includes officers from at least a dozen police departments, from large organizations like the Seattle police and Pierce County Sheriff’s Department to smaller ones like the Richland and Marysville police. Officers on the listserv are encouraged to adopt a Fight Club-style directive that precludes group members from discussing the existence of the list publicly. One document explicitly says: “Do not mention FITlist in your reports or search warrant affidavits.”
“In the context of face surveillance, a lot of access actually happens ad hoc, instead of through a written agreement or policy.”
Shankar Narayan, Director of the Technology and Liberty Project for ACLU of Washington, says that the use of a listserv to make backroom requests “with no opportunity for the public to know and to understand” what information is being shared is concerning. Narayan adds that secret partnerships like these could also be used to circumvent jurisdictions that have banned facial recognition technology.
As facial recognition technology has become more popular across law enforcement organizations, so has the backchannel sharing of related databases. In California, for example, local police departments share biometric images with other local departments, the state department of justice, other states in the region, police departments across the country, and the FBI. Earlier this year, OneZero revealed that despite public claims by New Orleans officials that the city’s police do not employ facial recognition, its officers employed a workaround, tapping state police to do the job.
“In the context of face surveillance, a lot of access actually happens ad hoc, instead of through a written agreement or policy,” Narayan says. “So an agency without a policy may get access [to facial recognition technology] through these requests.”
The emails obtained from FITlist provide a unique, firsthand look at how such secretive networks operate and how they may sidestep regulatory efforts meant to curb the use of facial recognition technology.
“We’re seeing jurisdictions like San Francisco and Somerville, Massachusetts enact bans and moratoriums on police use of facial recognition, and yet a list like this could allow those same agencies to circumvent those rules and access the technology,” Narayan says.
The documents obtained by OneZero suggest a pattern of backroom requests and resource sharing outside public scrutiny. One police department, in particular, the Pierce County Sheriff’s Department, was frequently tapped to conduct facial recognition searches on behalf of other jurisdictions either through FITlist or directly to Pierce County Sheriff Forensic Investigations Manager Steve Wilkins.
In the email exchanges with other law enforcement, Wilkins demonstrates that he was well-versed in conducting facial recognition searches and was ready to assist other departments. In his correspondences, Wilkins explains how he edits photos and weighs in on the correct angle and aspect ratio needed for a photo to work properly with his software.
One email exchange obtained by OneZero shows detectives from the Fircrest Police Department reaching out to Wilkins for facial recognition help on a vehicle breaking and entering case.
In the first email, a Fircrest officer writes a fellow detective in the department with the request that he send a surveillance photo “downtown” to be “matched” with a specific suspect whose name is provided in the email. “I know its [sic] him but I want to make curtain [sic],” the officer writes.
It is unclear whether “downtown” functions as a codeword for Wilkins, but the Fircrest detective follows up by forwarding the inquiry to Wilkins. The detective requests that Wilkins compare the surveillance photo to the person’s booking photo in his facial recognition software “to see if there’s a match.”
A couple of days later, Wilkins writes back to the detective telling him he “worked this several ways” but does not think that it is the suspect.
The incident raises two red flags: A potential lack of accountability across departments, and that Wilkins is attempting to match surveillance footage to a known suspect. Using facial recognition software to attempt to match an image to a known person is different than the way other police departments have said they use their programs. Police from other cities around the country have said that the standard way they use facial recognition is by taking images and running them through a program that compares them to a database of mugshots or driver’s license photos.
Asked about this investigation, Wilkins told OneZero, “When I am given a surveillance image and a possible suspect name I do not use the facial recognition system.” He added, “In this case, I would do a one-to-one comparison of the surveillance images with the known images of the suspect in the mugshot database.”
Wilkins said that in all instances before he’ll run a search in his facial recognition software, there needs to be “some sort of probable cause associated with the surveillance for any work to be done.”
One of the Fircrest detectives on the email chain, Robert Deal, responded to questions about these messages, telling OneZero: “I have been to Steve Wilkins’ presentation on facial recognition and have used him as a source several times in my investigations.”
Deal said the Fircrest Police Department does not have a written policy on facial recognition.
The trove of messages also provides fascinating insight into how police around the state collaborate to investigate crimes. The emails include straightforward requests from officers who attached a surveillance image, Facebook photo, or driver’s license, asking his or her fellow police officers if they recognized the suspect. In other messages, officers ask their peer group if they think fingerprint analysis or a facial recognition search would be helpful in particular cases.
On its face, the FITlist can be viewed as an example of a more collaborative, cross-departmental approach to solving crime. However, some transparency advocates raise concerns about the list’s secretive nature.
“This is a very troubling practice,” said Brian Hofer, executive director for non-profit Secure Justice, an organization that has helped write several of the bills adopted by cities that banned facial recognition. “If accurate, it demonstrates a clear intent to escape judicial oversight, public scrutiny, and disclosure laws. If you have to hide what you are doing, it’s probably not a good thing.”
Lawmakers will have to wait until at least 2020 to take another crack at regulating how facial recognition and other biometric surveillance are used to track Washington residents.
The number of cities banning the use of facial recognition around the country continues to grow. Lawmakers in Massachusetts are currently debating a statewide moratorium on the use of the technology. Legislators in California recently passed a new law that bans the use of facial recognition on police body cameras. At the federal level, congressional lawmakers are considering a number of bills that could limit the use of facial recognition. A coalition of more than 30 companies and technology trade associations is urging Congress not to ban the technology outright.
In Washington state, momentum to reign in the use of this technology has slowed. Earlier this year, a controversial data privacy bill failed to pass at the state level. This means lawmakers will have to wait until at least 2020 to take another crack at regulating how facial recognition and other biometric surveillance are used to track and monitor Washington residents. This also means that police can continue to use these technological tools in ways that are largely unchecked.
OneZero reached out to a number of Washington state police departments whose officers participate in FITlist, but none responded to a request for comment. It is unclear how long FITlist has been around for, or why it’s necessary to keep it secret.