The Digitization of the Carceral State: The Troubling Narrative Around Police Usage Of Facial Recognition Technology
Sara E. Yates*
Print Version: The Digitization of the Carceral State- The Troubling Narrative Around Police Usage Of Facial Recognition Technology
The technological veil conceals the reproduction of inequality and enslavement.
This Note applies a racial social control frame to the problem of facial recognition technology (FRT), showing how this technology may entrench preexisting inequalities and disparate treatment of people of color by law enforcement. Police usage of FRT will likely increase the targeted surveillance and investigations of marginalized communities, while being perceived as more objective or progressive than traditional police surveillance. In fact, the common argument in favor of FRT states that these tools could be effective devices of criminal justice reform if we corrected for issues of accuracy and transparency. Specifically, that if we corrected for the inaccuracies with identifications of people of color, and made the underlying process more transparent, then these would be good tools for the police in their crime fighting function. This Note critiques this argument on the basis that policing is, and always has been, about racialized social control as opposed to a purely crime-fighting institution. Therefore, these tools, and the progressive critique that they would be “good” if they had improvements in accuracy and transparency, misses the mark of the real problem with their use by police. The problem is that police in the United States primarily serve to further racialized social control, and in allowing police to use these tools that make them more effective we are strengthening this social control under a technological veil of “accuracy” and “transparency.” In doing so, we run the risk of believing they can be tools of progressive criminal justice reform.
I. Police As Racialized Social Control 487
A. The Origins of Police as Actors of Racialized Social Control 487
B. The Modern Use of Proactive Policing as Racialized Social Control 490
II. The True Problem of Facial Recognition Tools 494
A. Predictive Policing and the Issue of Accuracy and Transparency 495
B. FRT as a New Means of Racialized Social Control 497
III. Avenues for Change/Reform 505
In 1919, the Bureau of Investigation’s general intelligence division—the precursor to the FBI—directly targeted a man by the name of Marcus Garvey. Garvey had committed no crimes that justified his surveillance, even J. Edgar Hoover stated that Garvey had not violated any federal laws. The sole basis of his targeted surveillance, and attempts at deportation by the Bureau, was his organization the Universal Negro Improvement Association—which the Bureau believed was “agitating the Negro movement.” The Bureau infiltrated the Association with agents and provocateurs for years in attempts to halt the Black activist group, and to find grounds to deport Garvey. Finally, four years later, the Justice Department pulled together a “dubious mail fraud charge” and convicted Garvey. At the same time, as their targeting and surveillance of Black civil rights groups, “white vigilantes, police and soldiers targeted Black communities . . . includ[ing] the Red Summer of 1919, the Tulsa massacre of 1921 and scores of lynching[s],” none of which received the focused surveillance and targeting as the civil rights groups.
Likewise, throughout the civil rights movement in the 1960s the FBI used similar tactics to undermine and disrupt civil rights leaders. For example, the FBI’s Counter Intelligence Program (Cointelpro) targeted leaders like Martin Luther King Jr. and was designed to “[p]revent the rise of a ‘messiah’ who could unify and electrify the militant black nationalist movement,” as opposed to “prevent any violent acts they might perpetrate.”
In 2008, the FBI gained new investigatory powers through the Bush Administration’s authorization of an “assessment” type of investigation. Assessments did not require any factual basis of suspicion or wrongdoing in order to conduct very invasive investigations, including physical surveillance, commercial and governmental database searches, and employing informants to investigate targets. An FBI memo in 2009 justified an assessment investigation on Black populations in Georgia on the basis of “fears of a ‘Black Separatist’ terrorism threat.”
Most recently, we have seen upticks in surveillance resulting from the Black Lives Matter (BLM) protests in the summer of 2020. The rise of the BLM movement was in response to killings of Trayvon Martin in 2021, and subsequent trial and not-guilty verdict of his killer George Zimmerman, and the fatal police shooting of Michael Brown in 2014. Unsurprising after looking at the trends above, the FBI began an assessment investigation tracking BLM political activists for months. By 2017, the FBI created the “Black Identity Extremism movement,” which it characterized as a domestic terrorism category. Specifically, an FBI report stated that this terrorist movement was motivated by “perceptions of police brutality against African Americans,” and that “perceived unchallenged illegitimate actions of law enforcement will inspire premeditated attacks against law enforcement” by “Black identity extremists.” These investigations continued and took priority over more prevalent and violent white supremacist and militant groups. In looking just in these past 100 years, we can see that marginalized groups and communities of color are highly surveilled, and are usually harmed the most as a result of that surveillance.
All of these targeted surveillance, investigations, and arrests of people of color can be understood as methods racialized social control. In this Note, I apply a racial social control frame to the problem of facial recognition technology (FRT), showing how this technology may entrench preexisting inequalities and disparate treatment of people of color by the police. The basis for this argument is that law enforcement’s primary function is to establish and further racialized social control, and new technologies that provide police and law enforcement greater surveillance and investigatory powers will be used in furtherance of this function. This understanding is crucially important in the critique of new technologies and investigative tools, such as FRT, used by law enforcement. No matter the reforms put in to make FRT more “accurate” or its code more “transparent,” its use by law enforcement will create discriminatory and disparate impacts and effects on communities of color, because that is what law enforcement was created to do. FRT will just be more of the same type of racialized social control that has plagued communities of color, especially African Americans, since the founding of police in the United States.
This analysis of racialized social control and the problem of FRT proceeds as follows. Part I provides a portrait of the carceral state as it relates to police and their furtherance of racialized social control. In looking at police specifically, this Note shows the importance of understanding the proliferation of punitive responses to social inequality, and the furtherance of white supremacy. Part II explores predictive policing and FRT, and why the narrative that FRTs are problematic only due to their lack of transparency and inaccuracy is faulty. While these systems are not inherently biased and discriminatory, the way they have been created and designed has resulted in these tools reinforcing current social inequalities with respect to low-income individuals and people of color. Part III concludes by advocating for abolitionist approaches to the carceral state, including halting the use of FRT. I argue that acknowledging the crucial role that race, and white supremacy, has in the social control effectuated by the police—and the potential for FRT to be used to further this control—notions of reform based upon “transparency” and “accuracy” are ineffective to address the concerns with these tools and effective criminal justice reform.
In order to understand and critique the concerning use of FRT by the police, we must understand the broader framework of how police serve as actors of racialized social control. Since their inception, police have pursued practices and policies to maintain the role of white supremacy and control minority groups and have continued these types of practices through the present day. Potentially just as harmful is the lack of awareness of police’s function as actors of racialized social control, and instead the dominant myth that they are here to protect and serve.
Police in the United States originated from slave patrols in the nineteenth century. These patrols, and their related vagrancy-type laws, were a way to continue white supremacy and enforce racialized social control. Specifically, they were created to “prevent slave rebellions by enacting laws that prohibited slaves from traveling without a pass and permitted slave patrols to arrest slaves on mere suspicion of sedition.” After the Civil War and Emancipation, police were again used to maintain white supremacy and keep freed slaves from leaving the plantations through Black Codes. Specifically, “policemen in Southern towns continued to carry out those aspects of urban slave patrolling that seemed race-neutral but that in reality were applied selectively. Police saw that nightly curfews and vagrancy laws kept blacks off city streets, just as patrollers had done in colonial and antebellum eras.” The Supreme Court stated that these Codes established vagrancy laws that were “used . . . to keep former slaves in a state of quasi slavery.” These Codes’ vagrancy provisions swept up a wide range of conduct that was “designed to force the freed slaves to work for their former masters.” The laws were successful to that end, as freedmen were deterred from leaving the plantation for fear of getting arrested and charged under the Black Codes—furthering the system of racialized social control post-Civil War.
This type of social control was not just limited to the police in Southern cities but was also present in the North as well. In contrast to the South, the North had a different type of racialized social control because their economies were not dependent on the type of rural farming labor and its enslaved workforce, that was the backbone of the Southern economy. However, during the late 1800s and early 1900s, northern cities were facing an increasing amount of immigration resulting from the Industrialized Revolution. During this time, the Northern police—paralleling their Southern counterparts—organized around protecting the interests of the wealthy manufacturing industries and controlling the immigrants who were seen as racially and ethnically inferior. Northern police used their social control function against the poor, marginalized, and perceived ethnically inferior groups in similar ways as Southern police during Reconstruction.
For example, after the Civil War, Buffalo, New York, faced an increase of immigrants due to the economic opportunities stemming from industrialization. Specifically, Buffalo had a substantial influx of Polish immigrants that made up most of the unskilled labor and led to significant economic growth. The immigrants were herded to live in “undesirable living areas along the waterfront, around factories and near stock yards where they lived with little income in generally deplorable conditions.” Coincidentally, organized police proliferated at this time, with a quick uptick in growth from the 1870s until the 1900s.
The police in Buffalo were specifically created to serve and protect the interest of the white wealthy business owners who had “direct control over the police department.” One of the biggest motivations of the creation of the police in Buffalo was to quell and control the working class that was increasingly becoming dissatisfied with the wealthy business interests. As more immigrant workers entered the area, business interests started to become a greater priority over workers’ wages or their working conditions. The strikes of immigrants and other dissatisfied workers advocating for greater rights were deemed to be a “disorder” by the police that had to be quashed. The majority of this disorder was in the form of non-violent strikes, and at times the entire force would be deployed for the entire duration of the strike, no matter how long and no matter if it was non-violent. However, in quashing this disorder the police often used violent tactics and threats themselves to prevent these strikes and political assemblies of immigrant workers.
In parallel to its Southern counterpart, the Northern police in Buffalo made arrests as a way to control the labor issues and reinforce the white and wealthy social order, keeping the immigrants and marginalized communities as second-class citizens. Specifically, they made arrests for offenses such as “disorderly conduct, vagrancy, tramp, [and] drunkenness” as a way to “freely . . . control any group of working class people if the police [were] so inclined.”
This is but one example of how the Northern police used state violence as a way to aid in subordinance of marginalized groups across an array of axes, and also further the dominance of the socially powerful. The origins and historical use of the police, both in the North and South, illustrate that early police were not about crime-fighting. Instead, police were focused on maintaining the political, social, and economic interests of the dominant group at that time, while keeping the poor and racial and ethnic minorities in their place.
The racialized social control furthered by police has only continued into modern day America. Just as the Black Codes served to further racialized social control by the police, modern police practices, like broken-windows policing, has created the modern equivalent. The broken-windows theory of crime was expressed by social scientists James Wilson and George Kelling in 1982 and became one of the most prominent theories of crime and policing. This “quality-of-life” focused policing, is critiqued as being unresponsive to crime causes, facilitating police misconduct, and being highly discriminatory of poor communities of color, especially Black communities. Bernard Harcourt explains that under this theory there are two types of individuals, those that are law-abiding “families who care for their homes, mind each other’s children, and confidently frown on unwanted intruders” and those that are “disreputable or obstreperous or unpredictable people: panhandlers, drunks, addicts, rowdy teenagers, prostitutes, loiterers, the mentally disturbed.” Harcourt critiques the broken-window’s causation of crime—that minor social and physical disorder (broken-windows) creates an environment that fosters crime—and instead claims that “[t]o the contrary . . . the category of the disorderly is itself a reality produced by the method of policing.”
This social control method of policing that arrests the visibly lawless, or disorderly, for minor crimes creates a perception of the type of person that must be surveilled and monitored by law enforcement. Dorothy Roberts links this concept to the emergence of what she terms “Black criminality.” She states that the visible lawlessness emerges in two ways, first by “visible characteristics other than their criminal behavior. They look like criminals even when they are doing not more than standing still.” Second, she states that “[t]heir very presence on the street is considered harmful and must be eradicated.” The social control of the visibly lawless goes hand-in-hand with Black criminality, which is the culturally propagated stereotype and belief that “crime has a black face.” This belief is “deeply embedded in [an] American culture that is premised on the superiority of whites and inferiority of Blacks.” Not only does this racialized social control further embed the concept of Black criminality in the minds of the American public, but creates violations to civil liberties by creating second-class citizenry for individuals based on “presumed criminality [that] permits the state to minimize the rights of presumably lawless citizens while expanding the authority of presumably law-abiding ones.”
Similarly, a related critique became popular from Michelle Alexander’s illustrative and influential book The New Jim Crow in 2010. The New Jim Crow critique states that through the targeting of Black individuals, men specifically, through the War on Drugs, and their subsequent incarceration, the criminal justice system functions as a systemic and structural means of racial control analogous to the Jim Crow laws that were present from the 1860s through the 1960s. Much of this characterization is racialized and clearly evident in the framing of tough-on-crime policies, arrests and prosecutions. This targeting and racial profiling creates a feedback-loop because by increasing surveillance and targeting Black individuals, there is a corresponding increase in arrests and incarcerations, cycling back to the assumption that Blacks are more violent and commit more crimes. This is the self-fulfilling nature of increased Black surveillance and Black criminality. Furthermore, this is criminogenic and as police surveil and arrest Blacks for minor crimes—creating a criminal record—it creates a ratchet effect that can increase criminal penalties for more serious crimes.
The notion of Black criminality is actively used by police to justify stops and detentions. The most prevalent example of this was the stop-and-frisk policies in New York City. Stop-and-frisk was a broken-windows style of policing that disproportionally targeted Black and brown men for stops and searches. Young Black and brown men only make up around 5% of the population of New York, but between 2003 and 2013 made up 41% of the people stopped, and between 2014 and 2017 made up 38% of those stopped. In 2012, a class action suit was filed claiming that the practices were unconstitutional. The Plaintiffs argued that by the practice’s targeting of racial minorities the city and NYPD had violated the equal protection clause of the Fourteenth Amendment. Throughout the course of litigation numerous pieces of evidence showed the practice’s ineffectiveness at stopping crime. Furthermore, the city could not explain the massive racial discrepancies in the stops the officers chose to make. The District Judge posed an explanation herself stating that “the racial composition of the people stopped by the NYPD resembles what the NYPD perceives to be the racial composition of the criminal population because that is why they were stopped.” This explanation, describing Black criminality and the resulting racialized social control as the purpose of the program, was confirmed later in the case. During testimony, statements by New York City Police Commissioner Raymond Kelly discussing the apparent racial targeting of the program came to light. Kelly had stated that he targeted young men of color “because he wanted to instill fear in them, every time they leave their home, they could be stopped by the police.” The District Court held that the NYPD had been conducting unlawful racial discrimination against New York citizens, and unlawfully conducting searches and seizures.
From the slave patrols of the nineteenth century, to the Black Codes of Reconstruction, to the targeted surveillance of Black civil rights leaders and organizations from the 1910s up to the BLM movement in 2020, to modern methods of proactive policing including stop-and-frisk, law enforcement have continued their order-maintenance policies through their most recent use of predictive and investigative tools, including FRT.
The preceding part illustrated how police have historically been used against the poor and other dispossessed and marginalized groups. From the Southern police’s use of social control through the Black Codes, to the Northern police’s use of social control of immigrants, it is clear that crime-fighting was not the dominant purpose of police. On the contrary, police were created to maintain the political, social, and economic interests of the dominant group at that time, in addition to keeping poor, racial, and ethnic minorities in their place. Furthermore, as a source of hegemonic state violence, police used violence and threats of violence to subordinate and aid in the dominance of groups across axes of wealth, race, and ethnicity. Such historical underpinnings and foundations create significant doubt that modern technological tools that increase the police’s efficacy at their job would lead to desirable results.
One such tool, FRT, is an undesirable tool to give police, even if there are corrections for issues related to accuracy and transparency. The main critique of FRT on the basis of its accuracy and transparency misses the mark and the true issue with this tool. The police’s usage of FRT will likely increase racialized social control through targeted surveillance, investigations, and potential state violence against marginalized communities, while being perceived as more objective or progressive than traditional police surveillance.
This section discusses the recent growth of technological and predictive tools used by the police to illustrate the accuracy and transparency critique. This section further explores why, even if we solve for these two issues, FRT is not a tool that should be in the hands of law enforcement.
In recent decades, the United States has increased its carceral responses resulting in a surge in incarcerations as a part of its broader trend in punitive governance. Part of this is the increased use of actuarial tools by law enforcement. Predictive policing is such an actuarial tool; it provides “additional information about the places and persons involved in criminal activity that supplements, rather than replaces, existing police techniques and strategy.” As explained in the illuminative work, Against Prediction, Bernard E. Harcourt explained these systems as actuarial risk assessment tools, stating that “they are actuarial insofar as they use statistical methods . . . in order to determine the different levels of offending associated with a group or with one or more group traits, and based on those correlations, to predict . . . the past, present or future criminal behavior of a particular individual.” Predictive policing is broken down into either place-based policing or person-based policing. While both types of policing have almost always been present in our criminal justice system, they have undergone significant changes with the advent of more sophisticated technology and technological improvements in surveillance. Person-based predicative policing generally involves targeting individuals using criminal history to make predictions about which individuals are most at risk of violence. Person-based predictive policing creates many more concerns about racial bias and discrimination given the nature of the analysis and the lack of empirical support of its success—as compared to place-based predictive policing.
One of the main arguments in favor of predictive policing, and artificial intelligence (A.I.) tools like FRT, is that by using machine-learning instead of humans who are more prone to inaccuracies and biases, these tools “are in fact equal, or even less biased than the status quo.” Others have agreed to this point, arguing that because predictive policing lacks the subjective discretion that is present in traditional policing that it has the potential to be less discriminatory and thus is more objective. Furthermore, since it is more transparent than the subjective human mind of the officer, it can be examined and interrogated, leading to greater transparency and accountability. However, just because these systems are potentially capable of having increased transparency and accountability, this does not mean that these capabilities automatically follow.
Sociologist Ruha Benjamin expresses her concern that new technologies can perpetuate and “reproduce existing inequities but that are promoted and perceived as more objective or progressive than the discriminatory systems of a previous era,” calling it “the New Jim Code.” I have a similar concern as it relates to new technologies used by law enforcement specifically, no matter how “accurate” or “transparent” they are made to be. In looking at our current highly punitive and carceral approach to criminal justice, the goal of the state is not to adjudicate guilt but instead to control populations. These predictive tools are utilized to facilitate this goal and further racialized social control through surveillance, regulation, and punishment. Furthermore, predictive policing tools rely on data that is a result of racialized policing and arrests. And as a result, the outcomes are predetermined to correlate to past racial inequalities in policing, arrests, and convictions.
Furthermore, I have the concern that FRT is perceived as being a less biased and more progressive option than policing of the past, since certain human elements are removed from FRT. The issue with FRT is that it can carry a “scientific aura” that produces a mistaken belief of objectivity and an unjustified deference in its outcomes. This mistaken belief in a technological objectively allows for the continued presumption of Black criminality, and the resulting minimizing of rights of Blacks on the basis of predictive lawlessness. Similar to the loitering and vagrancy laws discussed in Part I, these tools can give police justification to arrest and surveil individuals on purely race-based suspicion and predicted lawlessness—distinct from any actual criminal conduct.
This is even more harmful than the types of visible lawlessness discussed in Part I, as this model of predictive lawlessness is much worse given the presumption of objectivity and increased deference to these tools and their outcomes. The police’s historical use of violence and threats of violence to subordinate groups across axes of wealth, race, and ethnicity, create significant doubt that FRT use by police would lead to desirable results, even if we correct for issues related to accuracy and transparency.
One of the biggest areas of growth for predictive policing has been with the increase of FRT in order to identify perpetrators. FRT is the “process of capturing and analyzing characteristics of a subject’s facial image input by sensors, and then comparing the captured image to a particular facial pattern or a collection of patterns stored in a database.” At its core, a facial recognition system entails three things, “(1) a mechanism comprised of sensors that capture the facial biometric data from the subject; (2) a mechanism to extract identifying features from the captured facial image; and (3) a matching methodology that processes and compares the presented facial data to reference data in order to make a recognition decision.” As stated above, two of the main critiques of the usage of FRT is based on its lack of accuracy—the fact that it misidentifies people of color and women at much higher rates that white males, creating discriminatory outcomes—and based on its lack of transparency—the issue that these have a black box issue and are mostly unrelated. These two critiques are not wrong, and they do raise important issues regarding risk-assessment tools, but they miss the mark on the real problem of FRT’s use by law enforcement. But before delving into the real issue of FRT, it is important to understand the above two critiques.
Beginning in the 1990s, the Department of Defense created a program called FRT, and since then it has proliferated across the public and private sectors. As early as the 2000s state and local governments began using FRTs, but they were considerably unreliable. However, they are becoming far more sophisticated as determined by the National Institute of Standards and Technologies (NIST), which conducts voluntary tests of these vendors every four years. And in 2010, NIST found that the accuracy rates had improved at an incredible pace, with its rates increasing tenfold between each round of testing. Even so, the FRTs that are used by law enforcement are not required to endure testing to determine effectiveness, accuracy, or system bias before being utilized against citizens. In fact, facial recognition is less accurate than fingerprinting, especially when it is used on a large dataset or used in real time to identify persons.
Furthermore, many of these systems show disturbing accuracy differences across gender, race, and other demographics—crucially that these systems are more likely to misidentify African Americans over any other race. In a 2012 study conducted by a senior FBI technologist evaluating the commercial vendor Cognitec—which was used by police in numerous states including California, Maryland and Pennsylvania—found that the facial-recognition programs performed “5-to-10 percent worse on African Americans than on Caucasians. One algorithm, which failed to identify the right person in 1 out of 10 encounters with Caucasian subjects, failed nearly twice as often when the photo was of an African American.” This discrepancy is further compounded by fact that African Americans are disproportionately targeted by police surveillance—up to 2.5 times more likely to be the targets of increased surveillance.
States and local governments aren’t the only ones using FRTs. The Federal Bureau of Investigation (FBI) has been using its own facial recognition programs for over twenty years. In September of 2014, the FBI stated that its Next Generation Identification (NGI) program was fully functioning, and within this NGI program was the new Interstate Photo System (IPS) that employs facial recognition onto a new and upgraded photo database. However, as soon as it was released, IPS faced serious backlash from privacy groups because the system mixed mug shot photos with noncriminal individuals (their photos were taken from employment records and background check databases). At the time, the system was predicted to collect up to 52 million faces in its database. The FBI’s IPS program still raises the same privacy concerns today for its incorporation of noncriminal photos into its facial recognition database.
However, federal law provides the FBI with significant authority to implement these programs and to maintain and collect identifications and criminal records. Even worse, despite contrary statements to Congress, the FBI “has not audited use of its face recognition system . . . . Only nine of 52 agencies (17%) indicated that they log and audit their officers’ face recognition searches for improper use.” Furthermore, the Georgetown Law Center on Privacy and Technology after combining state and local systems with FBI data, found that “law enforcement face recognition affects over 117 million American adults. It is also unregulated. A few agencies have instituted meaningful protections to prevent the misuse of the technology. In many more cases, it is out of control.” More concerning is that the Government Accountability Office stated that 64 million of those noncriminal American citizens have their faces imputed into the system without any notice or permission from their driver’s license photos.
One vendor, Clearview A.I., has emerged at the forefront of FRT, and has created mass controversy in its wake for its collection and utilization of digital biometric data of innocent Americans. Clearview A.I. was created in 2016, and according to the company more than 600 law enforcement agencies across the U.S. have started using Clearview since 2019, including local cops, the FBI, and the Department of Homeland Security. The system has a database of over three billion images that it claims to have “scraped from Facebook, YouTube, Venmo and millions of other websites . . . go[ing] far beyond anything ever constructed by the United States government or Silicon Valley giants.” The New York Times coverage of the vendor has catapulted the company into the light, with many—include huge platforms like Facebook and Twitter—up in arms about how Clearview acquired these photos.
Even so, federal and state law enforcement have stated that while they only have limited knowledge of the workings of Clearview, they had used the software to “help solve shoplifting, identity theft, credit card fraud, murder and child sexual exploitation cases.” In 2017, Indiana State Police started using Clearview and “solved a case within 20 minutes . . . . Two men had gotten into a fight in a park, and it ended when one shot the other in the stomach. A bystander recorded the crime on a phone, so the police had a still of the gunman’s face to run through Clearview’s app.” However, like most of these systems, Clearview has not been assessed by any independent studies. Georgetown Law Center on Privacy and Technology expressed its concerns over the accuracy of Clearview, stating that “[w]e have no data to suggest this tool is accurate . . . [t]he larger the database, the larger the risk of misidentification because of the doppelgänger effect. They’re talking about a massive database of random people they’ve found on the internet.”
All this attention has led to major concerns from the general public, state legislatures, and the federal government regarding Clearview’s practices. These concerns have manifested in litigation, cease-and-desist letters, and demands from major platforms such as Facebook, Instagram, LinkedIn, and YouTube to stop Clearview from scraping their data. On February 6, 2020, a spokesperson from Facebook stated that “[s]craping people’s information violates our policies . . . . Which is why we’ve demanded that Clearview stop accessing or using information from Facebook or Instagram.” LinkedIn has gone one step further and sent Clearview cease-and-desist letters, following in the steps of Twitter and YouTube who sent their own letters in January 2020, stating “[w]e are sending a cease & desist letter to Clearview. The scraping of member information is not allowed under our terms of service and we take action to protect our members.”
Senator Edward Markey, D-Mass., has been aggressively requesting answers from Clearview regarding its privacy and civil liberty infringements on American citizens. In a January 2020 letter, Senator Markey conveyed that the company’s FRT has “an alarming potential to impinge on the public’s civil liberties and privacy.” He further stated that the widespread use of Clearview’s FRT “could facilitate dangerous behavior and could effectively destroy individuals’ ability to go about their daily lives anonymously.” The Senator specifically asked Clearview to list all of the law enforcement or intelligence agencies that they either sold or marketed to, and how the company tests for accuracy and what safeguards are in place to protect individuals’ data. After receiving a response filled with “dubious claims” from Clearview’s CEO, Senator Markey sent another letter in March of 2020, stating the initial responses were unacceptable.
On January 27, 2020, a coalition of forty civil liberties and privacy organizations requested that President Trump suspend the use of FRTs across the Federal Government. The letter also addressed Clearview’s pervasiveness in law enforcement—with over 600 law enforcement agencies using Clearview in 2019—to demonstrate the need for action. The letter concluded that given the lack of accuracy and the significant bias and discrimination issues of these FRTs, a moratorium should be place on their use. The ACLU might even be suggesting that if the accuracy and transparency issues could be resolved they would be open to changing their view on the ban and moratorium if proper safeguards are put in place.
All of this is not to say that reforms addressing accuracy and transparency are not worthwhile, because they are, however it’s more important to look at how these are actually being utilized by law enforcement. How police use these tools is arguably more important to their potential discriminatory impact, than the individual tools themselves. The problem with focusing on problems of accuracy and transparency in FRT is that when those are fixed, we are left with a tool that may seem to be more “accurate” and “objective” and thus a strong assumption can be made that it will create “just” and “objective” results in practice. I do not believe this will be the case. This is because the actors who utilize these tools do not create “just” and “objective” results in practice. As demonstrated throughout Part I, law enforcement’s purpose is to maintain white supremacy and racialized social control through the policing of marginalized groups and communities of color. In fact, giving them tools that allow them to this job more efficiently, and with an added level of unjustified deference given its perceived objectiveness, is actually more harmful to criminal justice reform than leaving in the accuracy and transparency problems.
In looking at how police use FRT in practice, it appears that most are used for “low level crimes” and more minor investigations. Police and advocates of FRT would say that the ability to increase their investigatory powers is a good thing. A Cincinnati police detective stated that “[w]e try to use [FRT] as much as we can.” There is an argument that, on its face, FRT is an acceptable tool because FRT is really about investigations and solving crime, as opposed to order-maintenance and racialized policing. Furthermore, one could argue that its use as a crime-solving tool is very different than the policing of race, class, and space seen in broken-window style policing like stop-and-frisk. Moreover, the argument would continue and say that even if policing is primarily about racialized social control, FRT is just different than the racialized policing that operated against marginalized groups and communities of color in the past.
In looking more in depth at how these tools are used, it does not appear that FRT is sufficiently different from policing in general for its racialized social control purpose. First, the majority of crimes it is used on are minor quality-of-life crimes, as compared to more serious violent crimes. This is an important distinction because these quality-of-life crimes, and their broken-windows theory underpinnings, are highly racially discriminatory. Furthermore, quality-of-life policing’s purposeful racialized application has been clearly established. In New York, where stop-and-frisk was most prominently used, the NYPD used FRT more than 8,000 times in 2019, and overwhelmingly against people of color. Importantly, FRTs are being employed in very similar ways to BLM protests and leaders as targeted police surveillance was employed to civil rights groups in the 20th century.
In 2015, and again in 2020, the FBI used its state-of-the-art spy surveillance plane to monitor BLM protests. The FBI has stated that it’s committed to public safety and that its efforts are “focused on identifying, investigating, and disrupting individuals that are inciting violence and engaging in criminal activity. The FBI respects those who are exercising their First Amendment rights, including the right to peacefully protest.” However, the use of the surveillance spy plane presents a different justification. First, the plane is normally used for federal gang and drug busts, not for protests or other First Amendment activity. In fact, when members of Congress demanded that the FBI stop using FRT on BLM protesters the FBI defended its actions saying that it is focused on public safety and “does not conduct surveillance based solely on First Amendment protected activity.” Nathan Wessler, an ACLU attorney, stated that “[i]t’s now been well documented that a number of federal agencies wildly overreacted to protests in DC in deeply troubling ways . . . [and] [t]o learn that the FBI deployed its state-of-the-art surveillance plane to watch these historic protests raises additional troubling questions.”
Even more concerning, there are reports that police departments are contacting the FBI and requesting images or videos that link protesters to violent acts. The request of this information of peaceful protesters, as opposed to conservative militant groups, raises questions regarding the motivations of the police. In looking back on how law enforcement behaved during the 1960s with civil rights groups, the racialized social control motivation becomes more apparent. This is the way that law enforcement and the police operate, and have since their inception. There is no reason to assume that they will use these tools in ways that will not create racial discrimination and further racialized social control. This was apparent in the targeted surveillance of BLM protests in 2020. If these tools are allowed to be used by law enforcement, whether they have been reformed to address the accuracy and transparency issues discussed above, they will still be used disproportionally against marginalized groups and people of color, reinforcing notions of Black criminality, and creating justifications for further police actions on the basis of a more “objective” tool.
As described in the above Parts, I do not believe that FRT should be reformed to solve the issues concerning their use by law enforcement. The true problem, of racialized social control, is a far more nuanced and complicated problem to tackle—and one that cannot be fixed with one single solution. In fact, a range of solutions and interventions will likely be key to the problem expressed here, and to effective criminal justice reform more broadly.
The most straightforward solution is a full ban or moratorium on usage of FRT by law enforcement. This reform has been suggested by the ACLU and has gained traction with several states. California, New Hampshire, and Oregon all have prohibitions on the use of FRT by police, with California passing a three-year ban in 2019. In 2020, Congress introduced the Facial Recognition and Biometric Technology Moratorium Act of 2020, which would stop the federal use and funding of FRTs. The bill was partially in response to the wrongful arrest of Robert Williams, a Black man, and the ACLU’s subsequent complaint against the Detroit police department. The ACLU claimed that Williams was arrested due to “flawed face recognition technology” and as a result was detained for thirty hours.
Additionally, nineteen state legislatures considered measures to limit or place a moratorium on the use of FRT by law enforcement and in police body cams. Additionally, huge tech companies such as Microsoft, Amazon, and IBM have limited their use of FRT and have stated that they will not sell their FRTs to police departments until federal regulation is in place. While an effective reform at stopping the use of FRT, a ban fails to address the fundamental issue at play by its use, and also fails to create a reform framework that can be applied to other predictive and investigative tools used by law enforcement. Any reform to these systems must question what these tools are doing for the criminal justice system, and how they are being used in practice.
As this Note has demonstrated, the greatest harm from these systems does not come from these tools themselves, but instead from the unjust institutions that use them. The Note suggests reforms that adopt both an abolitionist approach and also focuses on antiracist practices. These reforms would have the goal of reducing the carceral state, including the reduction of police. While this may seem relatively straightforward, such reforms face significant difficulty due to the pervasiveness of the issue and “[b]ecause police have played the role of enforcing social control of poor people of color, it will be difficult to reorient this aspect of policing.” Furthermore, the racialized social control and furtherance of the white supremacy function of policing, makes it crucial that these reforms contain antiracist practices. As shown throughout the history of policing, “presumed criminality permits the state to minimize the rights of presumably lawless citizens while expanding the authority of presumably law-abiding ones.”  Aggressively policing people of color and marginalized communities were a means to protect and further white interests and political power. This is to say, that white Americans “are willing to tolerate intolerable amounts of state violence against black people because their white racial privilege protects them from experiencing this violence themselves and because they see this violence as necessary to protect their own privileged racial status.”
Even so, after the BLM protests in the summer of 2020 and the resulting mainstream attention to violence perpetrated by law enforcement, this is the time to push through these reforms. And that is what states are doing. In late 2020, Denver, Colorado, replaced police with health care workers for certain mental health calls including substance abuse. As of February 6, 2021, the program has been very successful in its six months of operation. The Denver police chief stated that this change “saves lives . . . [and] prevents tragedies” as there have been no arrests made by the workers “despite responding to hundreds of calls.” There is hope that the federal government will start to implement such programs as well. During Biden’s presidential campaign he did state that he “would fund initiatives to pair police departments with mental health professionals, substance use disorder experts, social workers and disability advocates.”
In combination with abolitionist reforms and a reduction in police, some have suggested the use of predictive tools themselves as reform tools. Similar to discussion above regarding accuracy, one argument is to turn the assessment tools inward to remove racially discriminatory factors, as it presents less difficulty than removing bias from human actors. This is suggested on the basis that “[i]t is arguably easier to remove race-based criteria from a computer algorithm than from a decision made by an individual actor, since algorithms could theoretically be programmed to discount any factors which have been tainted by prior racial discrimination.” However, the most meaningful abolitionist use of these technologies has to address the factors, proxies, and racialized data that has resulted from centuries of racialized policing. These technologies would have to work in ways that they currently do not and find ways to do more than predict outcomes based on past inequality. In understanding how the digitization of the carceral state has reinforced racialized social control, white supremacy, and past inequalities we can begin substantial reform work to create a criminal justice system that is just for everyone.
While a worthy, and seemingly effective reform, bans on FRT fail to address the fundamental issue of FRT and its use by the police and law enforcement. This Note has argued that the true issue does not lie with accuracy or transparency—as the ACLU stressed—but instead with the police’s function as actors of racialized social control. No matter how perfect the tool itself is, if the actor that wields it is created to maintain white supremacy and current social hierarchies, then the tool will never be used in a just way. This is a systemic and fundamental issue with how our police and law enforcement agencies were created to operate. It is inescapable that from these institutions’ inceptions they have acted primarily to reinforce social hierarchies while keeping people of color tied to notions of lawlessness and violence, requiring greater surveillance and interactions with the criminal justice system to keep them in their place. In order to create any kind of meaningful criminal justice reform, there must be an acknowledgement of the true function of these institutions. Furthermore, there must be implemented active antiracist and abolitionist reforms to start the process of dismantling the racialized social control furthered by the law enforcement in the United States.
- * J.D. Candidate, 2021, University of Colorado Law School; Lead Articles Editor, Colorado Technology Law Journal. I give my deepest thanks to all the individuals that assisted in the making of this Note. I thank Professor Benjamin Levin for his invaluable guidance and feedback throughout the evolution of this Note, and his continual encouragement and support. I thank all the members and editors of the Colorado Technology Law Journal that have helped me edit and refine this Note. And finally, I thank my family who have always supported me and pushed me to think critically about the world we live in. ↑
- . One-Dimensional Man: Studies in the Ideology of Advanced Industrial Society 32 (1964). ↑
- . Michael German, The FBI Targets a New Generation of Black Activists, The Guardian (June 26, 2020), https://www.theguardian.com/commentisfree/2020/jun/26/fbi-black-activism-protests-history. ↑
- . Id. ↑
- . See id. ↑
- . See id. ↑
- . See id. ↑
- . Id. ↑
- . Id. ↑
- . Id. (Furthermore, “one stated goal of the Cointelpro program was to inspire fear among activists by convincing them that an FBI agent lurked behind every mailbox.”). ↑
- . Id. ↑
- . Id. ↑
- . See generally Michael J. Klarman, From Jim Crow to Civil Rights: The Supreme Court and The Struggle for Racial Equality (2004); State v. Zimmerman, No. 12-CF-1083-A (Fla. Cir. Ct. July 13, 2013); Josh Hafner, How Michael Brown’s death, two years ago, pushed #BlackLivesMatter into a movement, USA Today (Aug. 8, 2016, 7:50 PM), https://www.usatoday.com/story/news/nation-now/2016/08/08/how-michael-browns-death-two-years-ago-pushed-blacklivesmatter-into-movement/88424366/ [https://perma.cc/CCQ3-55S6]. ↑
- . German, supra note 2. ↑
- . Id. ↑
- . Id. (describing that the surveillance of the Black civil rights groups took priority over investigations concerning mass shootings, including that of a Pittsburg synagogue). ↑
- . Alfred Ng, Facial recognition has always troubled people of color. Everyone should listen, CNET (June 12, 2020, 5:00 AM), https://www.cnet.com/news/facial-recognition-has-always-troubled-people-of-color-everyone-should-listen/ [https://perma.cc/SE78-FHQB]. ↑
- . Tracey L. Meares, Policing: A Public Good Gone Bad, Boston Rev. (Aug. 1, 2017), https://bostonreview.net/law-justice/tracey-l-meares-policing-public-good-gone-bad [https://perma.cc/33LB-JM5U] (citing Mychal Denzel Smith stating “In 1996, James Baldwin wrote . . . , ‘the police are simply the hired enemies of this population. They are present to keep the Negro in his place and to protect white business interest, and they have no other function.’ This remains as true today as it was in 1966, only now we have bought into the myth of police ‘serving and protecting’ wholesale.”). ↑
- . Vagrancy laws were characterized by two features: first, they were very ambiguous and therefore gave police lots of discretion in their enforcement, and second, they allowed for someone to be arrested even though they had not done anything criminal. Given this, “vagrancy law[s] [were] often the go-to response against anyone who threatened, as many described it during vagrancy laws’ heyday, to move ‘out of place’ socially, culturally, politically, racially, sexually, economically, or spatially…[and] [a]s one Supreme Court justice would write in 1965, vagrancy-related laws made it legal to stand on a street corner ‘only at the whim of any police officer.’” Risa Goluboff, The Forgotten Law that Gave Police Nearly Unlimited Power, Time (Feb. 1, 2016, 11:00 AM), https://time.com/4199924/vagrancy-law-history/ [https://perma.cc/762W-JLFH]. ↑
- . Dorothy E. Roberts, Foreword: Race, Vagueness and the Social Meaning of Order-Maintenance Policing, 89 J. Crim. L. & Criminology 775, 788 (1999). ↑
- . Id. ↑
- . Id. ↑
- . Sally E. Hadden, Slave Patrols: Law and Violence in Virginia and the Carolinas 219 (2001). ↑
- . City of Chicago v. Morales, 119 S. Ct. 1849, 1858 n. 20 (1999). ↑
- . See Gary Stewart, Black Codes and Broken Windows: The Legacy of Racial Hegemony in Anti-Gang Civil Injunctions, 107 Yale L.J. 2249, 2259–60 (stating also that “[t]hese typically harsh codes were indented to regulate the black community and ensure that the South would remain a ‘white man’s country’.”). ↑
- . Id. at 2259. ↑
- . Charles Hirschman & Elizabeth Mogford, Immigration and the American Industrial Revolution From 1880 to 1920, 38 Soc. Sci. Res. 897, 898 (2009) (“From 1880 to 1920, the number of foreign born [migrants] increased from almost 7 million to a little under 14 million . . . Immigrants . . . were concentrated in the rapidly growing cities of the Northeast and Midwest during the age of industrialization.”). ↑
- . See Matthew DeMichele, Policing Protest Events: The Great Strike of 1877 and WTO Protests of 1999, 33 Am. J. Crim. Just. 1, 6 (2008); see generally Sidney L. Harring, Policing A Class Society: The Experience of American Cities, 1865-1915 (2nd ed. 2017). ↑
- . Sidney L. Harring & Lorraine M. McMullin, The Buffalo Police 1872-1900: Labor Unrest, Political Power and the Creation of the Police Institution, 4 Crime & Soc. Just. 5, 6 (1975). ↑
- . Id. (“Forty to sixty people crowded into ten room houses and sometimes boarders could not rent homes but only spaces on beds for a night or day turn. Ninety-four percent of the Poles in Buffalo had an income of less than $635, the living wage for that time.”). ↑
- . Id. at 6–7. ↑
- . Id. at 10. “During the period under consideration virtually every major business interest in Buffalo was represented among the Police Commissioners at one time or another. . . . All major and many extremely minor decisions affecting the police were made by the two commissioners.” Id. at 8. ↑
- . See id. ↑
- . Id. at 10. ↑
- . Id. ↑
- . Id. at 12–13. ↑
- . Id. at 12. ↑
- . As stated by criminologists Hurbert Williams and Patrick Murphy “[t]he fact that the legal order not only countenanced but sustained slavery, segregation, and discrimination for most of our nation’s history and the fact that the police were bound to uphold that order set a pattern for police behavior and attitudes toward minority communities that has persisted until the present day. That pattern includes the idea that minorities have fewer civil rights, that the task of the police is to keep them under control, and that the police have little responsibility for protecting them from crime within their communities.” Hubert Williams & Patrick V. Murphy, The Evolving Strategy of Police: A Minority View, 13 Nat’l Inst. of Just. 1, 2 (1990). ↑
- . This style of crime-control is also known as “order-maintenance,” “zero-tolerance,” or “quality-of-life” policing. Bernard E. Harcourt & Jens Ludwig, Broken Windows: New Evidence from New York City and a Five-City Social Experiment 12 (Univ. of Chi. Pub. L. & Legal Theory, Working Paper No. 93, 2005). ↑
- . Id. at 3; Wilson and Kelling’s premise was that “minor social disorder—littering, loitering, public drinking, panhandling, and prostitution—as well as physical disorder—graffiti, abandoned buildings, and littered sidewalks—if tolerated in a neighborhood, produce an environment that is likely to attract crime.” Id. at 10. ↑
- . Shankar Vedantam et al., How A Theory of Crime And Policing Was Born, And Went Terribly Wrong, NPR (Nov. 1, 2016, 12:00 AM), https://www.npr.org/2016/11/01/500104506/broken-windows-policing-and-the-origins-of-stop-and-frisk-and-how-it-went-wrong [https://perma.cc/AZX8-6Y43]. ↑
- . Bernard Harcourt, Reflecting on the Subject: A Critique of the Social Influence Conception of Deterrence, the Broken Windows Theory, and Order-Maintenance Policing New York Style, 97 Mich. L. Rev. 291, 297 (1998). ↑
- . Id. ↑
- . Roberts, supra note 19, at 803. ↑
- . Id. (emphasis in original). ↑
- . Id. (emphasis in original). ↑
- . Id. at 805. (Citing to Jody Armour “it is unrealistic to dispute the depressing conclusion that, for many Americans, crime has a black face.” Jody D. Armour, Race Ipsa Loquitur: Of Reasonable Racists, Intelligent Bayesians, and Involuntary Negrophobes, 46 Stan. L. Rev. 781, 787 (1994). A 1990 University of Chicago study found that “over 56 percent of Americans consciously believe that blacks tend to be ‘violence prone.’” Id. (citing Tom W. Smith, Ethnic Images 9, 16 (Dec. 1990) (General Social Survey Topical Report No. 19))). ↑
- . Id. at 805. ↑
- . Id. at 811. ↑
- . See Aya Gruber, The Feminist War on Crime, 7 (U.C. Press, 2020) (“The Black Lives Matter movement and books like The New Jim Crow did much to publicize the endemically racialized nature of policing, prosecution, and punishment.”). ↑
- . See generally Michelle Alexander, The New Jim Crow: Mass Incarceration in the Age of Colorblindness (The New Press, 2010); James Forman Jr., Racial Critiques of Mass Incarceration: Beyond The New Jim Crow, 87 N.Y.U. L. Rev. 21, 27 (2012) (quoting Michelle Alexander “Quite belatedly, I came to see that mass incarceration in the United States had, in fact, emerged as a stunningly comprehensive and well-disguised system of racialized social control that functions in a manner strikingly similar to Jim Crow.”). The first Jim Crow laws can be traced to the Black Codes established in the wake of the Civil War in 1865. Jim Crow Laws, History (Feb. 28, 2018), https://www.history.com/topics/early-20th-century-us/jim-crow-laws [https://perma.cc/J4X5-V23Z] (last updated Feb. 15, 2021). ↑
- . See Elizabeth Hinton et al., An Unjust Burden: The Disparate Treatment of Black Americans in the Criminal Justice System, Vera Inst. of Just. 2–4 (May 2018), https://www.vera.org/downloads/publications/for-the-record-unjust-burden-racial-disparities.pdf [https://perma.cc/2S25-XJKS] (Racial disparities in arrests, prosecutions and incarcerations are “rooted in a history of oppression and discriminatory decision making that have deliberately targeted black people and help create an inaccurate picture of crime that deceptively links them with criminality . . . They are compounded by the racial biases that research has shown to exist in individual actors across the criminal justice system—from police and prosecutors to judges and juries—that lead to disproportionate levels of stops, searches, arrests, and pretrial detention for black people, as well as harsher plea bargaining and sentencing outcomes compared to similarly situated white people.”). ↑
- . See Roberts, supra note 19, at 818. ↑
- . See id. ↑
- . See David Cole, No Equal Justice: Race and Class in the American Criminal Justice System 16–62 (1999) (discussing racial bias in consent searches, pretext stops, quality-of-life policing, and drug courier profiles); see also Randall Kennedy, Race, Crime, and the Law 136–67 (1997) (criticizing racial discrimination in investigative policing). ↑
- . According to the ACLU of New York “innocent New Yorkers have been subjected to police stops and street interrogations more than 5 million times since 2002, and that Black and Latinx communities continue to be the overwhelming target of these tactics. At the height of stop-and-frisk in 2011 under the Bloomberg administration, over 685,000 people were stopped. Nearly 9 out of 10 stopped-and-frisked New Yorkers have been completely innocent.” NYACLU, Stop-And-Frisk Data, https://www.nyclu.org/en/stop-and-frisk-data [https://perma.cc/4XAN-8CHD]. ↑
- . Id. ↑
- . Michelle Shames & Simon McCormack, Stop and Frisks Plummeted Under New York Mayor Bill de Blasio, but Racial Disparities Haven’t Budged, ACLU (Mar. 14, 2019, 4:00 PM), https://www.aclu.org/blog/criminal-law-reform/reforming-police/stop-and-frisks-plummeted-under-new-york-mayor-bill-de [https://perma.cc/X6HS-HB27]. ↑
- . Floyd v. City of New York, 959 F. Supp. 2d 540 (S.D.N.Y. 2013). ↑
- . Id. at 556. ↑
- . Id. at 559. ↑
- . Id. at 586. ↑
- . Id. ↑
- . Id. at 606 (quoting 4/1 Tr. at 1589). See also supra note 9, as a comparison regarding the language used to justify surveillance of racial minorities (discussing the FBI’s Cointelpro program’s purpose as having “one stated goal . . . to inspire fear among activists by convincing them that an FBI agent lurked behind every mailbox.”). ↑
- . Floyd, 959 F. Supp. 2d. at 586; The Second Circuit stayed the decision and remanded it to be heard by a different Judge on the basis of bias on the part of Judge Scheindlin; however, in 2014 the City stated that it would drop its appeal, thereby effectively reinstating Judge Scheindlin’s order. Ligon v. City of New York, 736 F.3d 118 (2d Cir. 2013), vacated in part by Ligon v. City of New York, 743 F.3d 362 (2d Cir. 2014). ↑
- . See Alexander, supra note 50; See also Jed S. Rakoff, Why Innocent People Plead Guilty, The New York Review (Nov. 20, 2014), https://www.nybooks.com/articles/2014/11/20/why-innocent-people-plead-guilty/ (The former District Judge discusses in part the religious origins of the U.S. stating “Americans are notoriously prone to making moral judgements . . . [and] a by-product of this moralizing tendency is a punitiveness that I think is not likely to change in the near future.”). ↑
- . Andrew Guthrie Ferguson, Policing Predictive Policing, 94 Wash. Univ. L. Rev. 1115, 1130 (2017) [hereinafter Ferguson, Policing Predictive Policing]; see also, e.g., Andrew Guthrie Ferguson, The Rise of Big Data Policing (2017). ↑
- . Bernard E. Harcourt, Against Prediction: Sentencing, Policing, and Punishing in an Actuarial Age 10 (Univ. of Chi. Pub. L. & Legal Theory, Working Paper No. 94, 2005) (emphasis added). ↑
- . Andrew Guthrie Ferguson, Illuminating Black Data Policing, 15 Ohio St. J. of Crim. L. 503, 504 (2018) [hereinafter Ferguson, Illuminating Black Data Policing]. ↑
- . Ferguson, Policing Predictive Policing, supra note 66, at 1148. ↑
- . Ferguson, Illuminating Black Data Policing, supra note 68, at 524. ↑
- . Sarah Brayne, Big Data Surveillance: The Case of Policing, 82 Am. Socio. Rev. 977, 977 (2017); Sandra G. Mayson, Bias In,Bias Out, 128 Yale L.J. 2218, 2279 (2019). ↑
- . Mayson, supra note 71. ↑
- . Brayne, supra note 71. ↑
- . Ruha Benjamin, Race After Technology, 5–6 (2019). ↑
- . See Issa Kohler-Hausmann, Misdemeanorland 4–5 (2018). ↑
- . Jennifer L. Skeem & Christopher T. Lowenkamp, Risk, Race, and Recidivism: Predictive Bias and Disparate Impact, 54 Criminology 680, 682–83 (2016). ↑
- . See Elizabeth E. Joh, Feeding the Machine: Policing, Crime Data, and Algorithms, 26 Wm. & Mary Bill of Rts. J. 287, 289, 301 (2017); see supra Part I. ↑
- . A risk assessment tool called the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) was used in Florida on more than 7,000 arrested individuals with extreme racial disparities on black and white arrested defendants. It was reported that the tool incorrectly designated black defendants as twice as likely to be future criminals in comparison to the white defendants, and incorrectly designated white defendants more often as being low risk of being future criminals than the black defendants. Julia Angwin et al., Machine Bias, ProPublica (May 23, 2016), https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing [https://perma.cc/E2JG-SRBS]. ↑
- . Mayson, supra note 71, at 2280. ↑
- . Mohammed Osman & Edward Imwinkelried, Facial Recognition Systems, 50 Crim. L. Bull. 3, at I (2014). ↑
- . Id. ↑
- . Id. ↑
- . Clare Garvie & Jonathan Frankle, Facial-Recognition Software Might Have a Racial Bias Problem, The Atlantic (Apr. 7, 2016), https://www.theatlantic.com/technology/archive/2016/04/the-underlying-bias-of-facial-recognition-systems/476991/ [https://perma.cc/R6LU-CJZ7]. ↑
- . Id. ↑
- . Id. ↑
- . Id. ↑
- . Clare Garvie, Alvaro Bedoya, & Jonathan Frankle, The Perpetual Line-Up, Geo. L. Ctr. on Priv. & Tech. (Oct. 18, 2016), https://www.perpetuallineup.org/ [https://perma.cc/YW6D-42GV]. ↑
- . Garvie & Frankle, supra note 83. ↑
- . Id. ↑
- . See id. ↑
- . Press Release, Fed. Bureau of Investigation, FBI Announces Full Operational Capability Of The Next Generation Sys. (Sept. 15, 2014), available at http://www.fbi.gov/news/pressrel/press-releases/fbi-announces-full-operational- capability-of-the-next-generation-identification-system [https://perma.cc/GR9Z-92N3]. ↑
- . Russell Brandom, The FBI just finished building its facial recognition system, THE VERGE (Sept. 15, 2014, 10:51 AM), http://www.theverge.com/2014/9/15/6152185/the-fbi-just-finished-building-its-facial-recognition-system [https://perma.cc/8XCM-SGUF]. ↑
- . Id.; More recently the IPS program has 30 million faces in its database. Mariko Hirose, Privacy in Public Spaces: The Reasonable Expectation of Privacy Against the Dragnet Use of Facial Recognition Technology, 49 Conn. L. Rev. 1591, 1599 (2017). ↑
- . 28 U.S.C. § 534. ↑
- . Garvie, Bedoya, & Frankle, supra note 87. ↑
- . Id. ↑
- . Id. (“16 states let the FBI use face recognition technology to compare the faces of suspected criminals to their driver’s license and ID photos, creating a virtual line-up of their state residents.”). ↑
- . Kashmir Hill, The Secretive Company That Might End Privacy as We Know It, N.Y. Times (Jan. 18, 2020), https://www.nytimes.com/2020/01/18/technology/clearview-privacy-facial-recognition.html [https://perma.cc/CUG2-UAEX] (last updated Jan. 31, 2020). ↑
- . Id. ↑
- . Jon Porter, Facebook and LinkedIn are latest to demand Clearview stop scraping images for facial recognition tech, The Verge, https://www.theverge.com/2020/2/6/21126063/facebook-clearview-ai-image-scraping-facial-recognition-database-terms-of-service-twitter-youtube [https://perma.cc/UB32-5FV4] (last updated Feb. 6, 2020, 3:22 PM). ↑
- . Hill, supra note 98. ↑
- . Id. ↑
- . Id. ↑
- . Id. ↑
- . Porter, supra note 100. ↑
- . Id. ↑
- . Henry Kenyon, Markey presses facial recognition company over privacy concerns, Congressional Quarterly (Jan. 23, 2020), 2020 WL 371551. ↑
- . Id. ↑
- . Id. ↑
- . Edward J. Markey, United States Senate, Letter to Mr. Hoan Ton-That, (Mar. 3, 2020) (available at https://www.markey.senate.gov/imo/media/doc/Markey%20Letter%20-%20Clearview%20II%203.3.20.pdf [https://perma.cc/5R9G-V7DQ]). ↑
- . Henry Kenyon, Privacy groups urge White House to ban use of facial recognition tech, Congressional Quarterly (Jan. 28, 2020), 2020 WL 427785. ↑
- . See id. ↑
- . Id. ↑
- . See Abdullah Hasan, 2019 Proved We Can Stop Face Recognition Surveillance, ACLU (Jan. 17, 2020), https://www.aclu.org/news/privacy-technology/2019-was-the-year-we-proved-face-recognition-surveillance-isnt-inevitable/ [https://perma.cc/5C3F-3SQX]. ↑
- . Alfred Ng, Police are using facial recognition for minor crimes because they can, CNET (Oct. 24, 2020, 5:00 AM), https://www.cnet.com/news/police-are-using-facial-recognition-for-minor-crimes-because-they-can/ [https://perma.cc/W3EJ-SQQA]. ↑
- . Jon Schuppe, How facial recognition became a routine policing tool in America, NBC News (May 11, 2019, 2:19 AM), https://www.nbcnews.com/news/us-news/how-facial-recognition-became-routine-policing-tool-america-n1004251 [https://perma.cc/LDK8-HZ6K]. ↑
- . Id. (Describing the successes the tool has had across the U.S. in intercepting crime stating “In Colorado, local investigators have foiled credit-card fraudsters, power-tool bandits and home-garage burglars and identified suspects in a shooting and a road-rage incident. In San Diego, officers snap pictures of suspicious people in the field who refuse to identify themselves. The technology has led to the capture of a serial robber in Indiana, a rapist in Pennsylvania, a car thief in Maine, robbery suspects in South Carolina, a sock thief in New York City and shoplifters in Washington County, Oregon. In southwestern Ohio, officers are dumping images from Crime Stoppers alerts into their newly acquired facial recognition system and solving all sorts of property crimes.”). ↑
- . See supra notes 115 and 116. ↑
- . See supra Part I. ↑
- . Id. ↑
- . Evan Selinger & Albert Fox Cahn, Did you protest recently? Your face might be in a database, The Guardian (July 17, 2020 6:27 AM), https://www.theguardian.com/commentisfree/2020/jul/17/protest-black-lives-matter-database [https://perma.cc/2REQ-FKBF]. ↑
- . Dominic-Madori Davis, FBI reportedly used top spy plane to monitor Black Lives Matter protests, Business Insider (June 21, 2020 2:52 PM), https://www.businessinsider.com/fbi-used-spy-plane-to-monitor-black-lives-matter-protests-2020-6 [https://perma.cc/HMX8-XLSP]. ↑
- . Id. ↑
- . Id. ↑
- . Rae Hodge, Capitol attack: FBI mum on facial recognition, Clearview AI searches spike, CNET (Jan. 12, 2021, 1:36 PM), https://www.cnet.com/news/capitol-attack-fbi-mum-on-facial-recognition-clearview-ai-searches-spike/ [https://perma.cc/R6CC-8GKW]. (emphasis added). ↑
- . Davis, supra note 122. ↑
- . See supra notes 8–13 and accompanying text. ↑
- . Hassan, supra note 114. ↑
- . Pam Greenberg, Spotlight: Facial Recognition Gaining Measured Acceptance, NCSL (Sept. 18, 2020), https://www.ncsl.org/research/telecommunications-and-information-technology/facial-recognition-gaining-measured-acceptance-magazine2020.aspx#:~:text=Laws%20in%20California%2C%20New%20Hampshire,on%20the%20practice%20last%20year [https://perma.cc/X27M-3GB5]. ↑
- . Press Release, Legislation Comes After Robert Williams Shares Story of Being Wrongfully Arrested Because of the Flawed Technology, ACLU (June 25, 2020), https://www.aclu.org/press-releases/aclu-comment-bill-stopping-face-recognition-surveillance [https://perma.cc/H8ZE-R6SL]. ↑
- . Id. ↑
- . Id. ↑
- . Greenberg, supra note 129. ↑
- . Jay Greene, Microsoft won’t sell police its facial-recognition technology, following similar moves by Amazon and IBM, Wash. Post (June 11, 2020, 12:30 PM), https://www.washingtonpost.com/technology/2020/06/11/microsoft-facial-recognition/ [https://perma.cc/E45C-THXV]. ↑
- . Scott Holmes, Resisting Arrest and Racism – the Crime of “Disrespect”, 85 UMKC L. Rev. 625, 667 (2017) (Even with the difficulty, Professor Holmes stresses that “it is important to strategically negotiate aspects and terms of this social control to make encounters safer and [fairer]. This might include minimizing contact between police and poor people of color by deprioritizing enforcement of laws which are used to target poor people of color. Ordinances, which deprioritize marijuana where there are racial disparities in marijuana enforcement, offer an example. Also, agencies which stop issuing charges for tail lights, tinted windows, and other non-safety traffic violations have the effect of reducing racially discriminatory stops.”). ↑
- . Roberts, supra note 19, at 811. ↑
- . Dorothy E. Roberts, Democratizing Criminal Law as an Abolitionist Project, 111 Nw. U. L. Rev. 1597, 1606 (2017). ↑
- . See programs in Oregon, Pennsylvania, and California. See What is CAHOOTS?, White Bird Clinic (Oct. 29, 2020), https://whitebirdclinic.org/what-is-cahoots/ [https://perma.cc/9GE9-BJBM] (describing Oregon’s CAHOOTS (Crisis Assistance Helping Out On The Streets) policy initiative that was implemented in 1989. The initiative “mobilizes two-person teams consisting of a medic . . . and a crisis worker who has substantial training and experience in the mental health field” as an alternative to a police response for certain non-violent crises to “ensure a non-violent resolution of crisis situations.”); Laura McCrystal, Philly has put a behavioral health specialist in its 911 call center amid calls for police reform, Phila. Inquirer (Oct. 9, 2020), https://www.inquirer.com/news/police-mental-health-911-kenney-reform-behavioral-20201009.html [https://perma.cc/US7K-2F6E] (describing Philadelphia’s policy initiative implementing behavioral health specialists into the police dispatch center. Mayor Jim Kennedy stated that “[t]he goal of the program is to safely deflect individuals with behavioral health needs away from the criminal justice system and into more appropriate behavior health care or social services in the community.”); Mayor London Breed Announces Roadmap for New Police Reforms, Off. of The Mayor: News Releases (June 11, 2020), https://sfmayor.org/article/mayor-london-breed-announces-roadmap-new-police-reforms [https://perma.cc/PBA2-DTRV] (describing San Francisco’s attempts to reform the police and police interactions with communities). ↑
- . Li Cohen, Health care workers replaced Denver cyops in handling hundreds of mental health and substance abuse cases — and officials say it saved lives, CBS News (Feb. 6, 2021, 12:04 AM), https://www.cbsnews.com/news/denver-health-professionals-replaced-cops-in-handling-hundreds-of-low-level-incidents-for-6-months-and-successfully-did-so-with-no-arrests/ [https://perma.cc/DE5X-FM5N] (“Under the Support Team Assisted Response (STAR) program, health care workers are dispatched in lieu of police when responding to incidents involving issues with mental health, poverty, homelessness or substance abuse. STAR providers only respond to incidents in which there is no evidence of criminal activity, disturbance, weapons, threats, violence, injuries or “serious” medical needs.”). ↑
- . Id. ↑
- . Id. ↑
- . Cohen, supra note 139. ↑
- . Ric Simmons, Big Data and Procedural Justice: Legitimizing Algorithms in the Criminal Justice System, 15 Ohio St. J. Crim. L. 573, 577 (2018). ↑