Abstract
Excerpted From: Ariam Kiflemariam, “Any Black Man Will Do”: A Transparency Framework for Eyewitness Identification in the Facial Recognition Technology Era, 15 Washington Journal of Social & Environmental Justice 153 (June, 2025) (448 Footnotes) (Full Document)
“Injustice anywhere is a threat to justice everywhere.”
About twenty years ago in western Washington, sixteen-year-old Cameron found himself in the crosshairs of an impromptu snowball fight with friends when he decided to take cover in the corner of a neighboring apartment complex. After an hour-long snow brawl, the high schoolers retired to the warmth of one of their apartments. They heard a knock. Someone opened the door, and a police officer appeared. He pointed straight at Cameron and curled his index finger in a summoning motion: “You, come here.”
Unbeknownst to Cameron, a tenant of the neighboring complex called the police alleging one of the boys had threatened to shoot him. In reality, the tenant--a middle-aged white man--had opened his door to yell at them, sort of like a cantankerous old man telling kids to get off his lawn. One of the boys pointed two fingers at the man and shouted, “Get your ass back in the house!” before leaving.
Confronted by an armed officer in a neighborhood too familiar with police misconduct, an oblivious Cameron asked, “What did I do?” The officer responded by pointing his taser at the teenager and pulled the trigger, but it jammed. Cameron took his chances and fled the apartment, but he was eventually caught and handcuffed. The officer brought the neighbor to confront the cuffed teenager and asked, “Is this the one who threatened you?” “Yes,” the disgruntled neighbor responded. Cameron was charged with harassment shortly thereafter.
Several months later, moments before trial, Cameron and the neighbor crossed paths in the court restroom. Not recognizing each other, they exchanged pleasantries. “I was coming out the bathroom, dude's coming in and, you know, gives you that little good old American greeting. Hey, how're you doing? How's your morning going? Oh, yeah, man, it's just cold this morning.”
At trial, the neighbor was called to the stand to testify as an eyewitness. “So, as I'm sitting in there, I think I had on like a, like plaid or something. [The prosecutor asked], 'So can you point to who it was?’ And he was like, 'Yeah, yeah, the gentleman there in plaid,”’ pointing to Cameron, who was sitting at counsel's table next to his public defender. When the witness was asked how he could remember, “He was like, 'I just remember his eyes, his cold, dark eyes.”’
I asked Cameron to show me a photo of the person who actually interacted with the neighbor months earlier--a childhood friend who later confessed to Cameron that it was he who shouted at the neighbor. Cameron--tall and slender with a dark complexion--looked nothing like his friend--a significantly shorter, broader boy with a much lighter complexion.
Based on the testimony of the man who could not even recognize him moments before trial, Cameron was convicted and sentenced to one month of incarceration. He was unable to secure apartments and jobs for years following his conviction. “'Any Black man will do.’ That was what it was. That's a case of 'any Black man would do.”’
Twenty years later, in the early 2020s, Randal Reid spent almost a week in jail for theft in a state he has never visited in his life. Nijeer Parks was arrested and jailed for shoplifting candy and attempting to hit a police officer with a car--except he was thirty minutes away from the location of the crime at the time it occurred. Alonzo Sawyer spent nine days in custody for assaulting a bus driver and stealing the driver's cell phone-- except Sawyer was fast asleep next to his wife during the incident.
All of these individuals have something in common: they are innocent Black people arrested after being misidentified. Separated by 20 years of policy reforms and technological advancements, none of them were able to escape the consequences of racial bias and poorly regulated policing. They are most certainly not alone, but because few mechanisms exist to exclude faulty identification evidence, and because most criminal cases resolve in plea agreements, it is impossible to quantify the extent of its role in false arrests. What we do know is that misidentifications--whether aided by FRT or not--disproportionately affect Black and brown people. The FRT-misidentification conundrum represents a modern manifestation of longstanding anti-Black bias in the criminal legal system that has yet to be addressed with a solution proportionate to the damage it inflicts.
The legal literature confronting issues arising out of the use of FRT in law enforcement has focused on urging courts to expand the meaning of Fourth Amendment “searches” to include government use of biometric technology like FRT; restricting government use of FRT through legislation; formulating possible causes of action against private companies that supply law enforcement with biometric data; encouraging trial courts to re-assess their approach to expert witness evidence involving the use of FRT; and exploring the tension between interests in fairness (de-biasing FRT by introducing more diverse facial images) and privacy (freedom from oversurveillance).
This article provides a fresh contribution to this important dialogue, first, by highlighting the dangers of hidden police reliance on FRT, and second, by providing a transparency framework that compels the disclosure of FRT use to identify defendants while tackling the well-known--yet largely unresolved-- risks and inequities inherent to eyewitness identification evidence more broadly. Using FRT to supplement eyewitness identification procedures does not resolve problems with eyewitness misidentifications. In fact, its use without the enforcement of stringent safeguards exacerbates longstanding issues with eyewitness identification. FRT highlights existing gaps in how legislators and courts use and address eyewitness identifications, and should refocus our attention to the root cause: nonuniform and unreliable eyewitness identifications.
Against the backdrop of FRT use involving racial minorities and cross-racial eyewitness identifications, this article proposes state legislatures should enact expansive automatic open discovery laws, looking to New York's Criminal Procedure Law, Article 245 as a model. Article 245 balances interests in criminal prosecution and individual rights by imposing timing requirements on both the state and defendant to produce discovery, and provides an expansive list of discoverable materials required from both sides. It also provides defendants with the necessary information to meaningfully weigh the benefits and costs of going to trial versus taking a plea deal. At a time where the vast majority of criminal charges resolve in plea agreements, these timing and content requirements are necessary to ensure the actualization of constitutional protections and the public's interests in transparent criminal prosecution and public safety.
This article also examines seminal cases addressing eyewitness identification procedures and evidence in New Jersey, Washington, and Oregon, illuminating both important judicial observations in how eyewitness evidence should be assessed by courts and gaps that can be narrowed by implementing a statutory scheme like New York's Article 245. Integrating Article 245 and science-informed judicial decision-making, this framework appropriately balances the interests of states, defendants, and the public, while advancing equitable justice in cases involving eyewitness evidence and the use of FRT.
Given FRT's known flaws and a lack of widespread, comprehensive regulation on FRT use in law enforcement, combined with decades of research highlighting weaknesses in human perception and memory, law enforcement and judges cannot reasonably rely on unscrutinized eyewitness identifications--including those tainted by faulty FRT use identify suspects without proper procedural safeguards. Yet, false identifications continue, and innocent people remain vulnerable to erroneous accusations, arrests, and convictions.
Part I provides a background on FRT, its use in law enforcement, and an overview of the problems associated with human use of AI programs and their implications in the present-day context. Part II presents an overview of eyewitness identification procedures commonly employed by police, the prevailing judicial standard applied to those procedures, and the gaps and internal inconsistencies it leaves unresolved. Part III examines three approaches to mitigating eyewitness misidentification as demonstrated in seminal state cases in New Jersey, Washington, and Oregon. These cases, and the standards they set forth, provide procedural blueprints that strike at the heart of the constitutional issue grounding this analysis: the weaknesses of human memory and perception, implicit bias, and the justice system's frequent reliance on poorly regulated eyewitness procedures to identify suspects and secure convictions. Part IV offers a proposal that, by integrating key holdings in the seminal cases reviewed with New York's open discovery law, would provide a framework to meaningfully mitigate the risk of constitutional deprivations involving eyewitness identifications. Adopting this approach would also importantly mitigate the harms of FRT inaccuracies and cross-racial misidentifications, and misidentifications in general, while balancing the interests of states, defendants, and the public.
This article is largely inspired by false arrests that do not result in trial convictions. While not always as disastrous as a more serious conviction and/or lengthy incarceration, arrests of innocent people perpetuate a destabilizing and stigmatizing strain on communities of color that is often overlooked. Arrests in and of themselves have many negative and concrete implications both inside and outside the criminal context. Once you have been arrested--wrongfully or not--your image and other identifiable information remains stored in law enforcement databases, and information collected relating to your case is disseminated to other noncriminal agencies. Consequences flowing from arrests include harm to communities targeted by public order policing strategies, pressure to accept plea bargains, time and money lost due to mandated court appearances or other procedures necessary to resolve an arrest and/or charge, and the revocation of probation or parole, among many others.
Ramifications outside the criminal system can be just as dire: immigration enforcement officials use arrests as a way to screen undocumented persons for deportation; public housing officials rely on arrest information to identify potential breaches of leases by low-income tenants, imposing the threat of eviction; employers may terminate employees who have been arrested, regardless of innocence; child protective services may intervene with parental custody in the event of arrest; potential foster care homes may be rejected on the basis of the arrest of a resident of the home; and juvenile arrests reported to schools can lead to stigma for arrested students.
The multifarious, domino-like impacts of arrests are even more distressing in the age of FRT, where the sophistication and capacity of real-time investigation, surveillance, and criminal risk assessment expands in sync with rapid technological advancement. Most importantly for this article's purposes, actors in the criminal justice system frequently rely on eyewitness identifications to secure arrests and convictions, and this reliance is being compounded--not mitigated--by the incorporation of FRT. With significant advances in technology, known weaknesses in eyewitness identification and FRT, and particular frailties in cross-racial eyewitness identification, this reliance must be moderated to prevent misidentifications and false arrests.
[. . .]
New York's CPL Article 245 provides a robust framework for addressing the challenges of eyewitness identification and the use of FRT in criminal proceedings. The law mandates comprehensive discovery, ensuring that both the prosecution and defense have access to all relevant information, including witness statements, visual evidence, and expert witness data. This expansive approach to discovery is coupled with mechanisms for challenging identification procedures and provides remedies for non-compliance, which are essential for a system that seeks to uncover the truth and protect the rights of all parties. By balancing the state's interest in prosecuting crimes, the defendant's right to due process, and the public's interest in safety and accurate justice, CPL 245 offers a model for other states to follow in their efforts to reform their own legal procedures.
The integration of legal precedents from State v. Henderson in New Jersey, State v. Derri and State v. Butler in Washington, and State v. Lawson in Oregon further enriches this framework by incorporating scientific research and addressing specific issues related to eyewitness identification and FRT. Henderson emphasizes the importance of considering both system and estimator variables, while also mandating enhanced jury instructions to educate jurors about the fallibility of eyewitness testimony. Derri requires trial courts to consider modern science when evaluating evidence, though it gives broad deference to trial court discretion. Butler highlights the challenges of cross-racial identifications and the need for a strong evidentiary foundation to support specific jury instructions, while Lawson shifts the burden to the prosecution to establish the admissibility of eyewitness evidence under state evidence rules. Together, these cases create a foundation for evaluating FRT-aided identifications by mandating transparency and a heightened level of scrutiny.
While New York's CPL Article 245 presents a comprehensive approach, potential limitations exist that may make it difficult for other states to adopt the scheme, including funding and infrastructure issues, as well as differences in legal frameworks and political considerations. Despite these obstacles, the framework laid out in CPL 245, along with the insights drawn from the cases in New Jersey, Washington, and Oregon, offers a pathway for states to create their own statutory schemes that emphasize transparency, balance competing interests, and promote more accurate justice in cases involving eyewitness evidence and the use of FRT.
This proposed framework provides a crucial lens for examining cases like those of Randal Reid, Nijeer Parks, and Alonzo Sawyer, all of whom were misidentified by FRT, as well as Cameron, who was misidentified before even cell phones were commonplace but after Brathwaite was decided. The comprehensive discovery mandates of CPL Article 245 would have helped to reveal the flaws in the identification processes that led to their wrongful arrests. The emphasis on system and estimator variables from Henderson is key to understanding the factors contributing to these misidentifications, such as suggestive identification procedures and cross-racial identifications, flawed algorithms, or poor image quality. Lawson's shift of the burden of proof to the prosecution would have compelled the state to demonstrate the reliability of the eyewitness evidence, while Butler underscores the danger of misidentification in cases involving cross-racial identifications, which are particularly prone to error. By implementing these reforms, states can work to ensure that the justice system does not perpetuate the kinds of errors that led to these individuals' wrongful arrests.
J.D. Candidate, 2025, University of Washington School of Law.
