Fiscally Effective With Disastrous Consequence: Race, Rights, and AI-Based Predictive Policing

I. Introduction

In the summer of 2020, over 1,400 mathematicians signed a letter urging colleagues to cease collaboration with police because of their racially discriminatory practices.1 Their critique focused on predictive policing, a law enforcement strategy that involves algorithmic data analysis to forecast where and when a crime will occur, plus the identities of both perpetrators and victims.2 It is employed by numerous law enforcement agencies across the country with the justification that it is cost-effective, objective, accurate, and overall more efficient. Predictive policing is nothing new, but the use of artificial intelligence (AI) in it is.3 AI relies on the data that it is fed, which is problematic because of the biases that pervade historical criminal data.4 Credible concerns regarding racial prejudice, and legality/constitutionality have consequently arisen. AI-driven predictive policing is inherently biased because of the data it is trained on, and is incompatible with the Fourth and Fourteenth Amendments.

II. Comparative Analysis Between Municipalities

The application of predictive policing systems varies between municipalities. This analysis will focus on three law enforcement agencies: the New York City Police Department (NYPD), the Los Angeles Police Department (LAPD), and the Chicago Police Department (CPD). The LAPD was among the first to adopt predictive policing, with Chicago and New York’s police departments following shortly after.5 These police departments “often turned to data analytics companies such as PredPol and Palantir” for software that could forecast crime.6

The LAPD’s main predictive policing programs, PredPol and Operation LASER, were discontinued due to public backlash against its reinforcement of racial discrimination. An internal audit of LASER was conducted in 2019 by the LAPD Inspector General, which revealed “significant inconsistencies” in how an individual was selected and kept in LASER’s system.7 The Latino and Black communities comprised 84 percent “of the 233 ‘active’ chronic offenders.8 While the review did not note this as an overrepresentation because it reflects violent crime arrest demographics between 2012 and 2018, these communities have historically been over-policed and subject to discriminatory laws prior to and during this timeframe.9 PredPol ended because of intense pressure, and financial limitations.10

The NYPD used several predictive policing software programs before switching to Patternizr, its in-house algorithm, in 2016.11 Though Patternizr does not use racial data, it is still capable of producing “racially-disparate results” due to implicit biases held by analysts.12 The data in the NYPD’s crime database likely overrepresents Latino and Black men because of biases in crime reporting and over-policing, increasing the likelihood of them being matched to crimes.13 Stop-and-frisks have consistently enabled racial bias by the NYPD. The overwhelming majority of those stopped are people of color. In 2024, 60 percent of stops involved Black people, and 31 percent involved Latinos.14 While officers are required to fill out a form detailing each stop, underreporting is a significant issue identified by the court-appointed Monitor, who ensures reforms are adequately implemented after the NYPD was again found liable for racial profiling in Floyd v. City of New York.15 This suggests that the data fed into the database is incomplete, and continues to reflect racial biases.16 This contributes to the feedback loop, intensifying existing racial prejudices within the data.

In 2012, Chicago piloted the Strategic Subject List (SSL), which compiled a list of individuals “considered most likely to commit gun violence or to be a victim of it.”17 Like New York City and Los Angeles, Chicago’s Black and Latino communities were listed at disproportionate rates compared to their white counterparts.18 The CPD imposed significant levels of surveillance on these communities in response, exacerbating the racial divide in arrests and reinforcing systemic racial biases.19 The program was suspended in 2019.20

The deployment of predictive policing programs in these cities has reinforced and amplified long-lasting systemic inequalities. These programs rely on data that is marred by racial discrimination within the criminal justice system, making objective likelihood difficult.

III. Legal & Constitutional Implications

The issue of predictive policing raises legal challenges relating to constitutional rights and protections, specifically those of the Fourth Amendment. When law enforcement relies on algorithms to identify potential suspects, questions arise regarding whether resulting predictions justify police action. The Fourth Amendment protects the people from unreasonable searches and seizures.21 Terry v. Ohio, argued in 1967, defined the scope of reasonable suspicion, establishing that officers must have “specific and articulable facts,” not subjective intuition.22 Predictive policing depends on broad historical data inputted, not precise information. Algorithmic output therefore is not specific enough to link an individual to a crime; it is merely a prediction. Additionally, algorithms frequently re-flag overpoliced neighborhoods as high-risk.23 This creates a feedback loop wherein the predictions are constantly used as justification, making these communities continuously vulnerable. Another relevant case is Leaders of a Beautiful Struggle v. Baltimore Police Department (BPD). BPD’s Aerial Investigation Research program used small planes to capture imagery of about 90 percent of Baltimore.24 The imagery was meant to track movements related to serious crimes, and would be “retained for forty-five days, while reports and associated imagery data were retained indefinitely as needed for legal proceedings.”25 The case was decided in favor of the plaintiff, setting the precedent that mass, suspiciousless surveillance by law enforcement agencies violates the Fourth Amendment when conducted warrantlessly. Predictive policing techniques resemble the surveillance strategies used by the BPD, and should be subject to the same judicial scrutiny.

The 14th Amendment guarantees equal protection, and fair legal treatment for all by state and local governments.26 The Equal Protection Clause is threatened by the use of historical crime data in predictive policing due to the bias it perpetuates against marginalized racial groups. This guarantee is violated when certain groups are disproportionately penalized due to systemic shortcomings created by the government. The due process clause is also undermined by predictive policing as nobody can be deprived of "life, liberty, or property, without due process of law."27 Using black-box algorithms (meaning: internal decision-making processes are hidden) to label individuals as threats subjects them to increased surveillance while freeing law enforcement of accountability.28 Defendants cannot challenge this designation, or evaluate its fairness.

IV. Conclusion

Predictive policing has repeatedly demonstrated the ability to reinforce racial bias, and undermine constitutionally protected rights. The data that predictive policing software is fed reflects decades of overpolicing and discriminatory criminal laws, continuing cycles that marginalize already disadvantaged racial groups.29 Established legal doctrine strongly suggests that predictive policing is unconstitutional, violating the 4th and 14th Amendments. The use of AI-driven predictive policing ought to be re-evaluated by municipalities, and regulated by the legislature.

______________________________________________________________________________ Footnotes Davide Castelvecchi, Mathematicians Urge Colleagues to Boycott Police Work in Wake of Killings, nature (June 19, 2020) [hereinafter Castelvecchi 2020] Castelvecchi 2020, supra note 1; Tim Lau, Predictive Policing Explained, Brennan Center for Justice (2020) [hereinafter Lau 2020] Pranshu Verma, The Never-Ending Quest to Predict Crime Using AI, The Washington Post (2022) Artificial Intelligence in Predictive Policing Issue Brief, NAACP (last visited Oct. 11, 2026) [hereinafter NAACP] Lau 2020, supra note 2 Verma 2022, supra note 3 Lau 2020, supra note 2; Grace Baek & Taylor Mooney, LAPD Not Giving up on Data-Driven Policing, Even After Scrapping Controversial Program, CBS News (2020) [hereinafter Baek & Mooney] Baek & Mooney, supra note 7 Baek & Mooney; Hinton, Elizabeth Hinton, LeShae Henderson & Cindy Reed,Vera Inst. Just., An Unjust Burden: The Disparate Treatment of Black Americans in the Criminal Justice System 1 (2018) Leila Miller, LAPD Will End Controversial Program That Aimed to Predict Where Crimes Would Occur, Los Angeles Times (2020) John McCormick, NYPD Built Bias Safeguards Into Pattern-Spotting AI System, Wall Street Journal (2019) Molly Griffard, A Bias-Free Predictive Policing Tool?: An Evaluation of the NYPD's Patternizer, 47 Fordham Urb.L.J., 43, 65 (2019) Id. at 66 Stop-and-Frisk Data, NYCLU (2025) Id.; Thirty-Third Report of the Independent Monitor (Corrected) at 4, Floyd v. The City of New York, No. 1:08-cv-01034-AT (S.D.N.Y. October 6, 2026) ECT 990-1; Floyd v. City of New York, 959 F. Supp. 2d 540, 645 (S.D.N.Y. 2013) Stop-and-Frisk Data, NYCLU, supra note 14 Lau 2020, supra note 2 Aaron Tucek, Constraining Big Brother: The Legal Deficiencies Surrounding Chicago’s Use of the Strategic Subject List, The University of Chicago Legal Forum (2018) Id. City of Chicago Office of Inspector General, Advisory Concerning The Chicago Police Department's Predictive Risk Models 2 (2020) U.S. Const. amend. 4 Terry v. Ohio, 392 Oyez 21 (1967) Selin Arslan, The Legal Implications of Predictive Policing Algorithms: Bias, Oversight, and Public Accountability, 2 Legal Stud. Digital Age 49, 50 (2023) Leaders of a Beautiful Struggle v. Baltimore Police Department, 135 Harv. L. Rev. 920, 920 (2022) Id. at 921 U.S. Const. amend. 14, § 1 Id. Elena Li, Predictive Policing and the Constitutional Category Error, Claremont Journal of Law and Public Policy (last visited Oct. 11, 2026) NAACP, supra note 4