Police body-worn cameras (BWCs) embody hope and promise for the objective preservation of criminal procedure evidence and an enhanced means for police to provide transparency and accountability regarding their activities and practices. Yet, BWCs can only operate to capture phenomena within optical and audio recording limits of the camera itself. This is a developing capability, for which standards for use are flexible. In the “real world,” BWC value can be limited by operator factors, departmental regulation (or lack thereof), and the same human observer effects and cognitive biases of any other visual evidence. With the advent of new technologies that can be incorporated into BWC hardware and systems—facial recognition and AI scrubbing, to name two—a critical analysis of the use of BWC evidence, its limitations, and its foibles is necessary.
February 20, 2024 Feature
Police Body Cameras: Analog Thinking about a Digital Question
Hon. Roderick Kennedy (Ret.)
There are enough aspects concerning the use of BWCs to make this article a virtual drop in a bucket. I hope to introduce some of the numerous issues that are involved in police departments’ use of BWCs and suggest a diversity of resources on the subject as well.
Police Body-Worn Camera Basics
BWCs are a component of the changing face of “evidence verité.” First used in Rialto, California, in 2012 and soon after in New York City, BWCs are now ubiquitous in the United States. BWCs are small enough to go where officers go, and, if activated, should record what transpires in officers’ presence. Rialto’s small pilot program correlated with an 88% decline in complaints against officers and a 60% reduction in use-of-force incidents. BWCs have been occasionally resisted by police officers and their unions for the potential to record non-incident-related behaviors.
Early on, BWCs were touted to be part of “best practices” in policing. The Department of Justice (DOJ) stated, “Body-worn cameras hold tremendous promise for enhancing transparency, promoting accountability, and advancing public safety for law enforcement officers and the communities they serve.” It suggested that BWCs are “thought to improve police conduct and transparency, especially regarding police use of force.” DOJ undertook a grant program to assist local/tribal law enforcement to purchase and deploy BWCs. By 2018, about 80% of large departments had BWCs. More than 80% had policies for their use, but the policies are not standardized, nor is their enforcement.
A BJA 2022 practice evaluation mentions BWCs’ ability to capture an “objective record” of events that does not force “agencies to rely solely on written reports from officers or accounts from citizens . . . potentially encouraging mutual accountability during encounters.” These are valid considerations, although skepticism of both the objectivity of the visual record, its value to objectivity and accountability, and how jurors use the evidence persist. The quality of policing, especially as it affects racial groups, revealed by the volume of videos of police interactions entering the public forum, has rightly been called into serious question. The evidence was often clear; its release to the public by police agencies varies from a trickle to a cataract, and, thus, its effect to encourage police accountability and thus provide a moderating effect on police/citizen encounters is uncertain. Studies have indicated a decrease in use of force and quality of police/citizen interaction, with citizens reciprocating; in another, BWC use correlated with an increase in police use of force and arrests.
For police, use of BWCs carries increased costs to departments of acquiring software, developing new procedures and policies for officers, increasing data storage, and hiring personnel to handle redaction, editing, retention, and review of footage. Prosecutors and defense attorneys must have access to review, categorize, and notate footage, adding cost to the justice system. BWC programs, while promising, are still a police instrument that is in flux.
Video Evidence Comes to Court
The impact of witnesses on trials has been studied since the nineteenth century, as evidenced in the venerable Wigmore on Evidence. Introducing photography, film, and video in court as evidence required recalibration by courts of the impact of how to present potentially graphic and disturbing evidence verité to lay fact finders. With video from citizen cell phones and BWCs, digital evidence has caused great acceleration in both the capabilities and dangers of contemporarily recorded images coming to court in the twenty-first century.
In 2016, Mary Fan pointed to courts’ relying on testimony to find the facts of a case and being a step away from the circumstances as to what happened for events they adjudicate, to a time where “the staple events of criminal procedure law will be recorded.” The 2014 death of Eric Garner in New York at the hands of police officers and the 2020 murder of George Floyd in Minneapolis were both vividly recorded by bystanders, uploaded on social media, broadcast by news media, and used in court. The impact on both judicial proceedings and public opinion is substantial as video of police/public interactions are both aired to the public and brought to court as evidence.
Mary Fan advocated for a reexamination of court rules to protect the integrity of the audiovisual record when offered as evidence, both to present a complete and contextualized picture and to actively provide a legal standard to protect against unquestioned assumptions on the part of fact finders that video presents unmediated truth. The basis: There is a difference between legally relevant truth and the depiction of it on video. Attention to “perceptual yardsticks” and providing interpretive tools for fact finders is a critical necessity—so is a detailed understanding of the limitations of the BWC medium. The ability to replay in court videos of things that witnesses previously only described brought a revolution in the capabilities of presenting evidence, but also the need for regulation and skepticism concerning its power. Why is this supposedly “objective” source of evidence drawing caveats and skepticism?
BWCs are increasingly employed in the field, while their capabilities are expanding to capture visual and other evidence beyond the “normal viewpoint” of a human observer, involving similar cognitive foibles. Courts and lawyers must guard against the potential for cognitive mischief in the minds of fact finders at trial.
Evaluating Fact Finders’ Accuracy Amid Cognitive Biases Regarding Video
Intuitively, a viewer can believe what they see on video is an objective record of events. The viewer has no more acuity than a dirty lens and can only observe for the length of the recording of a particular incident, all with no more perspective than the camera affords. Since events exist in a 360-degree world, lawyers must take care to recognize, identify, and explain to fact finders the different meanings that can be found in this visual evidence. This can be as much circumstantial as visual. For instance, what is portrayed in a video might be an interaction with a person with a mental disorder that is not recognized by the officer, who is facing a verbal onslaught and passive resistance that could be explained by the behavioral problem. Without understanding the circumstantial environment, the fact finder may be no better at assessing the evidence than the officer.
Fact finders may place overreliance on video evidence owing to its immediacy of presentation, its first-person appearance and image salience compared to written or verbal testimony, their preexisting biases toward video as “reality,” or its emotional content, compounding the “eyewitness” effect with credulity as to the nature of video evidence itself.
Neal Feigenson and Christina Spiesel’s work on the psychology of video evidence highlights common practices and problems with presenting unstaged, unedited footage of legally relevant events captured by a range of recording devices from BWCs and surveillance cameras to citizens’ cell phones. Combined with assumptions from day-to-day experience from the news, social media, and other sources, jurors may come to think and expect that video is likely to be reliable, probative, and persuasive, even while their perceptions and interpretations of this evidence are beholden to cognitive biases of which they are not aware. These can affect how jurors see, hear, and construe the evidence. But their views also can change in the course of trial. The ubiquity of raw BWC footage released in public by police agencies may exacerbate a tendency to desensitize the public to the nuance of legal and constitutional limitations on police behavior, and perhaps operate to dehumanize the subjects of police interactions owing to a process of seeing a never-ending panoply of worst behaviors without critical context by which to evaluate them.
Viewers of BWC video thus share many characteristics of eyewitnesses, and the same problems regarding ability to perceive and the existence of factors affecting what is “remembered” assume similar weight when applied to video from BWCs. Observers of evidentiary video are prone to “naïve realism,” a belief that the video shows them reality as it was. Given that many police videos depict emotionally charged situations on both sides, the emotional impact can increase observers’ vulnerability to many biases that foreclose recognizing their subjective involvement or fragmentation of perspective or enhance a tendency toward identification bias. Jurors are as susceptible to confirmation bias—using evidence to reinforce prior belief—as any human. In this reality, a viewer may become overconfident in their beliefs and judgments they derive from the video—prompting them to be less inclined to look for alternative explanations or be aware of how their conclusions are shaped by preconceptions.
As video conveys both less and more than appearances, lawyers need to take control of multiple meanings and educate judges and juries in ways that best serve their theories of the case.
Body-Worn Camera Videos: Effects of Police Policies and Procedures
Using BWCs is an aspect of police practice that must include comprehensive regulation by policies and procedures. Officer’s activation of cameras prior to engaging in interactions is of paramount importance. Failure in this regard is critical. BWCs have a buffer that records a period of time (about 30 seconds) prior to actual activation, without audio, but officers often fail to capture a complete incident, losing important evidence. Similarly, departments have problems with officers placing their cameras on their equipment belts sufficiently low as to obscure the field of view. And, inevitably, despite the high hopes and goals for cameras attached to officers producing objective and trustworthy primary evidence, police officers have been called to task for fabricating video evidence of drug discoveries. However, more prosaic manipulation of police video is not uncommon.
Here is a strong example of the limitation of police video: In 2014, deputies were pursuing Derrick Price from the scene of a SWAT activation. The close-in police body camera footage depicted deputies attempting to restrain Price, who was face down on a sidewalk. Officers were yelling for him to “stop resisting” while attempting to handcuff him. Much of the video faced away from the action. However, a surveillance video from the building in front of which the sidewalk was located showed the entire incident: Price running from deputies who were chasing him and, prior to their catching up, putting up his hands and passively laying on the sidewalk before the deputies proceeded to kick and punch him. Was the police video manipulated to face away from much of the beating? Where were videos from the other deputies? N.B.: The video also was not produced by the Sheriff’s Department for sixteen months after the incident, a phenomenon regarded as inimical to promoting departmental transparency and public trust.
As vividly seen in the Price video, the video suffered from late activation and a limited view, showing only the violent motion by body parts of officers who were yelling at Price to “quit resisting!” The same happens in audio content of BWC recordings in the author’s experience. Police officers commonly attempt to structure the narrative of a video in their on-camera actions and words, for example, by yelling for a subject to quit resisting in the absence of resistance. Oral statements made by police that are recorded on the video create new problems regarding hearsay and confrontation rights in evidence law. Behavioral changes by officers owing to being on camera are noted. Officers can be prone to play to the camera or narrate the video as it unfolds to supplement the images of the unfolding action. This attempt to influence a viewer’s perception of the action is worthy of skeptical judicial scrutiny, particularly given the credulity with which the video evidence can be received. How video influences officer narratives arises in the context of report writing as well.
Officer Access to Body-Worn Camera Video and Standards for Reporting Use of Force
Police officers write reports, which are often the first thing the prosecutor and defense attorney see. In most police departments, officers are generally allowed, if not encouraged, to review their video in the course of writing their reports. The ability to conform recollections to a virtual playback of the incident requires evaluation. For use-of-force (UOF) purposes, the extent to which access to data that they did not possess during the stress and distractions of an incident is later made available to restructure their narrative is an issue.
In the context of BWCs as being vital evaluative tools in UOF investigations, the existence of departmental and legal standards regarding what is reasonable or excessive force under the Constitution governs evaluations of UOF. Evaluating excessive force is judged by legal standards like Graham v. Connor, requiring assessment of officer perceptions of the need to use necessary, minimal, and proportional force to counteract an immediate threat at the time of the incident. The objective reasonableness of an officer’s subjective perceptions of an immediate deadly threat at the moment force is used is critical, not the arc of circumstance recorded in real time by a camera. Under Graham, “hindsight” is not relevant to judging the reasonableness of the force used. This requires assessing unfiltered recollections to evaluate officers’ actions and conduct at the time they used force, under the totality of those circumstances.
The availability of BWC video to officers creating their written reports and the timing of officer access to them are important concerns. The officer prepares an incident report at the end of shift and frequently prepares a separate report dealing specifically with the underlying basis for the UOF—a legal purpose reflecting the severity of the crime involved, the level of threat from the subject, and the officer’s subjective perception as to whether it was an imminent or immediate threat to the officer personally. Officers writing reports unaided by access to BWC video are removed from corroborative or contradictory “factual” information that may be present, which could allow for “hindsight”-based justifications and explanations that are impermissible under some department UOF policies and the Graham line of cases. Using BWCs for report writing should be explained and examined.
Last, it is not uncommon for BWC video or audio to be redacted by law enforcement agencies prior to release outside the organization. Redactions can include confidential information, graphic content, and information protected by privacy provisions.
Constitutional Considerations Increase as Body-worn camera Capabilities Increase
Fourth Amendment and privacy considerations become important as both the scope and particularity of information gathered by BWCs may exceed constitutional and other legal standards for evidence gathering. First Amendment rights may be chilled by incidental surveillance that algorithmic searches can pinpoint individuals for proper, improper, or mistaken purposes.
Existing models of thinking about police body cam video as impacting searches or seizures under the First and Fourth Amendments abound. Traditionally, the use of cameras that mimic the extent to which human vision can operate has been free of strong restrictions on their use and, to a fair degree, treated with some deference by the courts as presenting an image worthy of prima facie accuracy. Scott v. Harris represents a significant example. However, intrusion into areas subject to “reasonable expectations of privacy” may expand by BWCs’ ability to collect and visualize data outside the visible spectrum, as in the Kyllo case, or collateral data from the whole BWC field of view. There is nothing to prevent BWCs from acquiring and transmitting incidental data to which artificial intelligence can be applied to discern possibly relevant information.
BWCs are not limited to recording only images. A BWC also records audio; the microphone is nondirectional in its ability to gather audio or visual evidence that might be outside the constitutional requirement of particularity if what is to be searched were to become an issue. Conversations that are not relevant to the investigation but may contain content that on later listening shows itself to be otherwise germane to police purposes can be recorded. Juvenile victims or suspects and their statements may be outside the scope of proper recording and need special treatment. GPS capability allows police departments to collect information on locations where there are multiple incidents, or places particular persons who are not at the time of recording legitimate subjects for investigation may frequent.
The ability to interface BWC video with facial recognition and AI is also concerning. First, facial recognition systems have a recognized tendency to perform poorly with the faces of women, children, and persons with darker skin tones, leading to over-policing among vulnerable populations and increased danger of misidentifications. With the advent of real-time video from BWCs, the possibility of video dragnets or suspicionless investigations being undertaken in real time increases as another Fourth Amendment concern as images of persons not relevant to the officer’s immediate activities are flagged for other reasons by facial recognition technology—random persons recorded anywhere may have their data stored and recalled for other purposes later. The ability to identify persons involved with First Amendment–protected activities for later investigation can chill the right to associate and expose protesters to persistent surveillance and identification in police databases. AI video analytics is an algorithm-driven process that can isolate persons and objects appearing on the video, also identifying clothing, tattoos, and other objects, along with hair color, facial hair, and skin color, further impacting the right to associate or be free of suspicionless searches.
The continuing and advancing use of BWCs by police can affect a panoply of legal issues and areas of concern from civil rights through criminal procedure. There does not seem to be an aspect of it without some cause for concern, from the gathering of the data to its storage, retention, and use by the police agencies that use the cameras. The disparate landscape of regulation of body cameras—despite serious effort from organizations such as Police Executive Research Forum and the New York University Policing Project—deserves continued attention. This is particularly so as the capabilities for enhanced data acquisition and real-time analysis increase.