Body Cameras Should Not Be Live-Streamed
I recently wrote about the Fresno police department’s experiments with “threat scoring” software and why that approach to law enforcement is a big problem. One part of the Washington Post story on this that I didn’t mention is Fresno’s “Real Time Crime Center,” which is just the kind of centralized panoptic surveillance center that has always inspired our opposition to pervasive government surveillance camera networks. At the center, according to the Post, operators can pan and zoom “an array of roughly 200 police cameras perched across the city. They could dial up 800 more feeds from the city’s schools and traffic cameras,” and hope to add thousands more video streams “from local businesses that have surveillance systems.”
But one other part of Fresno’s vision for the center caught my eye. According to the Post, the city hopes “to add 400 more streams from cameras worn on officers’ bodies.”
Body camera video should NOT be used in systems that live-stream video to centralized monitoring stations and allow police management to activate and monitor any given officer’s camera at will. That is not the right balance between privacy and oversight and practicality. Body cameras are there to provide evidence and accountability in disputed circumstances, not to help furnish police departments with all-seeing views of American cities and towns.
Body camera manufacturers are starting to offer cameras with streaming functions. In some cases the focus is simply on using streaming as an alternative to docks as a way of uploading video to the cloud for storage, but the industry also seems to be moving toward selling the panoptic potential of this technology. This product, to take just one example, boasts that video from “thousands of officers can be managed in one central location.”
Congress has dedicated $7 billion toward creation of a dedicated first responder broadband wireless network called FirstNet, so police departments are going to have a lot of dedicated bandwidth that they could use for this streaming (as well as other potential surveillance applications as I discuss here).
Why do I think remote activation and monitoring is the wrong approach? It is true that despite our opposition to government video surveillance of the public, we at the ACLU have conditionally supported police body cameras. That’s because those cameras function to monitor the police, something that is sorely needed given the widespread problem of police abuse. It’s true that body cameras also function to monitor the public, but that is offset not only by the advantages the cameras offer in letting the public monitor its police forces, but also by the fact that the only members of the public who are recorded are those already in the presence of a police officer. We worry about the chilling effects of government-run video, but people in the presence of a police officer are usually already pretty chilled; anything they do that the police officer decides or alleges is illegal can lead to their arrest (and without video it will be only their word against the police officer’s that they are innocent, and their word is not likely to prevail). If done right, we think the advantages of police body cameras outweigh the potential surveillance downsides.
But the purpose of body cameras is to serve as an independent record when something dramatic goes down. It is not to feed video to centralized panoptic surveillance centers. While the privacy threat created by video streams from police body cameras is tempered by the chilling effects dynamic I just mentioned (the fact that the video only comes from scenes where a police officer is present), the streaming of bodycam video adds significantly to the power of centralized monitoring systems. Picture the situation in a city like New York, where there are 40,000 sworn police officers circulating around the city, disproportionately deployed in low-income areas and communities of color. The centralized live-streaming of body cameras would instantly super-charge the surveillance powers of the authorities, especially in communities that are already heavily policed. Simply put, this is too much power to place in the hands of anyone. It raises the prospect of abuse, and will create significant chilling effects.
True, thousands of camera feeds—or even a few dozen—would be more than any human could monitor, but they would be subject to the panoptic effect: when you don’t know whether or when you are being watched, you have to assume you’re being watched at all times. In addition, the AI technology for automated monitoring of video streams (the implications of which I wrote about here) is likely to continue improving, with the possible result that thousands of video streams could actually be meaningfully monitored.
Live monitoring significantly disrupts the careful controls and balances that are necessary if police body cameras are to strike the right balance between oversight and privacy. In the body camera programs we support, the vast majority of bodycam video never sees the light of day: it isn’t retained long, it isn’t subject to analytics, police officials do not plow through it, and it is not released to the public unless it depicts an officer’s use of force or felony arrest, or there has been a complaint against a police officer.
Remote activation and monitoring also poses a risk to the privacy of police officers, and of management misuse (such as looking for nickel & dime violations on the part of officers who are disfavored by management, a concern we first raised in our original 2013 white paper on bodycams). Last week I attended a daylong symposium on body cameras in Washington sponsored by the Police Foundation and other law enforcement organizations, and the subject of streaming cameras came up. Some of the police officers and officials present thought that rank and file police officers and their unions would resist this technology on privacy grounds.
Remote activation would also risk the privacy of the people the officers are with at the time; it could trump an officer's decision to deactivate a camera in sensitive situations in accord with good policies, and it’s unclear how citizens (or the officer for that matter) would receive notice that a camera has been remotely turned on. It would intensify the already real risk that cameras will create new barriers between police and citizens.
Finally, as cybersecurity expert Paul Rosenzweig pointed out at the symposium, the streaming of video would greatly increase the cybersecurity challenges posed by this technology—i.e. the threat of being hacked.
Of course, for any technology proponents can tell a story in which that tool saves the day. That is not the same thing as a technology possessing advantages that outweigh its disadvantages, or a technology actually proving effective in real-world practice.
Overall, the disadvantages of live-streaming, centrally monitored body camera video are not counterbalanced by significant improvements in oversight, which if done right should work well enough with after-the-fact video, and would not significantly improve with live streaming, at least in any way that I can see. Police departments should not go down this road.