The Department of Homeland Security’s 287(g) program authorizes US Immigration and Customs Enforcement (ICE) to delegate federal immigration authority to state and local police. Though the program has existed for 30 years, it has expanded rapidly at an unprecedented pace under the current administration, growing from fewer than 140 enrolled agencies at the end of 2024 to more than 1,730 by July 2026. More than 65 percent of 287(g) agreements use the task force, or “force multiplier” model, which integrates immigration enforcement into routine police work.
The presidential executive order driving expansion aims to remove people without legal status who have committed serious crimes, a goal the order describes as “critically important to national security and public safety.” The effort has required a significant increase in federal, state, and local law enforcement personnel and resources, potentially diverting them away from other public safety priorities.
While the scale and speed of 287(g) expansion underscore the need for rigorous research on the effects of this program, lessons from a broad body of policing and crime research suggest that the 287(g) program as currently structured and implemented may be unlikely to achieve its public safety goal—and may even undermine public safety in the process.
How 287(g) has changed local police involvement in immigration enforcement since 2025
The proportion of ICE arrests involved in the 287(g) program has risen notably, and the type of arrests they are used for has changed dramatically.
Beginning in 2022, between 80 and 100 percent of ICE arrests in a month involved people with criminal convictions or charges, but since 2025, this proportion has dropped to only half of all arrests.
Despite the stated goal of pursuing people with more-serious charges and convictions, the percentage of people with minor misdemeanor charges or convictions has risen, while the proportion with felony charges has declined. In other words, the arrest patterns do not follow the “worst of the worst” framing.
Why 287(g) ICE arrests have risen
Financial incentives can encourage a greater volume of arrests, creating a sort of quota system similar to those used in stop-and-frisk styles of policing. Such systems often prioritize the quantity of stops and detainments over the quality of each arrest’s contribution to public safety or justice.
Federal funding reinforces this dynamic. Under the 2025 budget reconciliation law, which vastly increased the budget for enforcement agencies, ICE reimburses participating agencies for officers’ salaries, benefits, and overtime, and pays quarterly performance awards tied to the number of people officers help locate.
Two potential outcomes of expanding 287(g) policing
Decades of research point to two possible consequences of this approach to policing:
- Ineffective law enforcement. A robust body of evidence suggests the most effective way to improve public safety is to focus on the small number of people and places that drive most serious violence. Focused deterrence, group violence reduction strategies, problem-oriented hot spots policing, and community violence intervention work by targeting those few “hot people in hot places,” as Thomas Abt writes in Bleeding Out. This work can only be done by local agencies and organizations that know their communities intimately.
However, the incentives created by ICE reimbursement for 287(g) participation encourage the opposite: a high rate of arrests for people who pose little public safety risk. Research shows broad, high-volume enforcement spreads finite resources thin and produces large numbers of low-yield contacts (e.g., resulting in no enforcement action or yielding negligible amounts of contraband), making it both expensive and ineffective. - Increased distrust in law enforcement. Indiscriminate, numbers-driven enforcement has a bad track record. New York’s stop-and-frisk approach produced hundreds of thousands of stops that mostly turned up nothing, fell disproportionately on Black and Latino residents, and was ultimately determined to be unconstitutional. This eroded community trust, itself an input to solving crime.
The 287(g) task force model sits in this history: It was shut down in 2012 after federal investigators found officers in Arizona and North Carolina using their immigration authority to profile Latino residents. Yet in September 2025, the US Supreme Court stayed an order (PDF) that had barred immigration stops based on factors like location, apparent ethnicity, type of work, and speaking Spanish or accented English.
There are already signs that community trust is eroding and fear of law enforcement is prevalent: The December 2025 Well-Being and Basic Needs Survey found that 21 percent of adults in immigrant families who were aware of local enforcement action in the previous year had chosen not to speak to police, and 17 percent had not driven a car because of immigration concerns. Research on past immigration enforcement initiatives also suggests such dynamics can increase victimization (PDF) without reducing crime.
How much safety, at what cost?
These lessons matter now because of the sheer scale of the resource commitment for expanding immigration enforcement. As enforcement grows, so do the trade-offs. Producing arrests at this volume has already required diverting large numbers of officers from other public safety work.
A Cato Institute analysis of ICE disclosure records estimates more than 25,000 officers have been diverted to immigration enforcement, including roughly 17,000 reassigned from other federal agencies—about one in five Federal Bureau of Investigation agents, half of the Drug Enforcement Administration, and more than two-thirds of the Bureau of Alcohol, Tobacco, Firearms, and Explosives—plus more than 8,500 state and local police newly deputized to make immigration arrests as of September 2025. Over the same period, criminal referrals from several of those federal agencies fell.
This forces the fundamental question: Is this trade-off improving public safety? Evidence from violence reduction research shows that a high volume of enforcement is not the same as safety, and that broad, indiscriminate enforcement can erode the very trust that safety depends on. If public safety is the objective, a large body of evidence points to the efficacy of using data to identify the small number of individuals or groups causing harm in a community and focusing intervention and enforcement efforts accordingly.
These types of targeted strategies are core among the efforts credited with contributing to historic declines in violence in cities nationwide that began in late 2022. With homicide caseloads falling, police agencies have an opportunity to invest in solving and clearing more violent crimes, which not only delivers justice to victims but also deters crime. That work depends on victims and witnesses willing to come forward, but enforcement tactics that undermine public trust and cooperation can impede this work.
Current federal policy is deviating from these best practices. As policymakers consider further expansion, they must not overlook the lessons of stop-and-frisk, past iterations of the 287(g) task force model, and decades of research on what reduces violence.
