Trump is Quietly Building a Deportation Army out of State and Local Agencies

Enrollment in the 287(g) program, which deputizes local police to do federal immigration enforcement, has exploded in recent weeks—but the administration has been surprisingly silent about it.

Since January 20, the Trump administration has aggressively—yet quietly—expanded a controversial program that delegates immigration enforcement powers to state and local law enforcement agencies across the country that are normally prohibited from investigating and arresting non-citizens based on immigration violations.

The 287(g) program, named for the corresponding section of the Immigration and Nationality Act added in 1996,¹ enables the Department of Homeland Security to “enter written agreement” with state and local agencies to “perform the function of an immigration officer.”² 287(g) agreements³ are uniquely powerful for two reasons: (1) they allow ICE to quickly and dramatically expand its enforcement reach far beyond its current formal capacity, and (2) they allow non-federal agencies to overcome the longstanding prohibition against states doing their own immigration enforcement. In practice, 287(g) creates (or at least aspires to create) a direct pipeline from routine interactions with local police to detention and deportation.

Federal immigration authorities did not sign the first 287(g) until after 9/11, when the Florida Department of Law Enforcement signed one in 2002. The number of agreements grew gradually after that, then shot up at the end of the Bush administration and beginning of the Obama administration.⁴ There are three key generalizations about 287(g) from this era:

1. Most—though certainly not all—of these 287(g) agreements were located in the U.S. South, in places like North Carolina and Georgia, which had seen significant relative growth in Latino residents in previous years⁵ and where I conducted much of my own research on 287(g) as a graduate student.

2. ICE developed two versions of the 287(g) program: the jail enforcement model,⁶ in which suspected noncitizens were screened for immigration violations at the point of book-in, and the task force model, the more controversial of the two, which gave officers the ability to investigate immigration status in the course of routine policing. ICE also created a quasi-third model that simply combined the first two.

3. Most 287(g) agreements were with county sheriffs, rather than other enforcement agencies, for two reasons. First, sheriffs are elected, which means that they had a political incentive to incorporate immigration enforcement into their campaign platforms by publicly pursuing 287(g) agreements.⁷ Second, sheriffs run the county jails where all law enforcement agencies in the county book people they arrest; thus, jails function as a strategic node for ICE in the larger enforcement system.

287(g) did not go unchallenged. The geographic expansion of immigration enforcement away from the border—what we call “interior enforcement”—was accompanied by a shift in immigrant rights organizing towards new direct and indirect anti-deportation activism (e.g., the New Sanctuary Movement),⁸ publicizing and litigating the harms caused by local anti-immigrant policies (e.g., the tragic story of Mark Lyttle), documenting concerns about the relationship between 287(g) and racial profiling, and the spread of sanctuary city policies that aimed to serve as a counterbalance to ICE’s infiltration of local jurisdictions.⁹

In 2012, after a slew of lawsuits and investigations called attention to major problems with 287(g), specifically with the aggressive task force model, the Obama administration terminated all task force agreements, but left the jail enforcement agreements largely in place through the end of his second term.

Then Trump happened. In one of his first executive orders in 2017, President Trump instructed the Department of Homeland Security to revitalize the 287(g) program by “empower[ing] State and local law enforcement agencies across the country to perform the functions of an immigration officer in the interior of the United States to the maximum extent permitted by law”¹⁰—language that Trump would use again in the executive order titled “Protecting the American People Against Invasion” issued on the first day of his second administration.

During Trump’s first administration, the number of 287(g) jail enforcement agreements doubled to about 80—a significant expansion to be sure, but jail enforcement agreements alone did not represent a historic expansion compared to the Obama years. Instead, the Trump administration, specifically Thomas Homan¹¹ as far as I can tell, invented a new 287(g) agreement to get around legal challenges that had made sheriffs nervous about getting involved in federal immigration enforcement. The Warrant Service Officer model of 287(g) more narrowly delegated to local law enforcement agencies the power to execute federal civil immigration warrants.

This alone contributed to a sudden spike in 287(g) agreements between the start of 2019 and the end of 2020, reaching over 150 active agreements at a single time and heavily concentrated in Florida but also in North Carolina, Georgia, and Texas. Unlike the early expansion of 287(g) during the Obama administration, the growth of 287(g) in 2019 and 2020 did not result in a directly observable growth in national deportations.¹² In either case, the administration did not draw much attention to the expansion of the 287(g) program and the enormity of other immigration policies at that time, including family separation and Title 42, meant that few people outside of specific policy circles paid much attention to 287(g).

In short, between 2010 and 2020, 287(g) went from being the central controversy of national immigration enforcement policy to being largely an afterthought.

On the campaign trail, Joe Biden emphasized his distance from Trump on immigration and even promised to end the 287(g) program as part of his package of immigration policy reforms. But once in office, his administration adopted a position of equilibrium: no new agreements would be signed and, with some rare exceptions, no old agreements would be terminated. Thus, the total number and type of 287(g) agreements remained largely static, with some slight but observable attrition over four years. And like the first Trump administration, 287(g) remained largely overshadowed by border policy.

And this, friends, is where I had left it. During the pandemic, I had worked on a project with a brilliant intern, Mario Marset, to document the history of 287(g) agreements over time. But due to our respective career changes, the primacy of other immigration crises, and the demanding ongoing validation that the project required, the 287(g) paper fell just short of publication.

Joel Sati, professor of law at the University of Oregon, came to my rescue when he invited me to participate in a ‘works in progress’ conference this past weekend in Eugene.¹³ I decided that this would be the ideal opportunity to get feedback on the paper and get it published, but I was intentionally avoiding adding any new data to the paper (since this was one thing holding me back). As the conference approached, Wendy Fry, whose work for CalMatters you should definitely follow, reached out to me about a story on 287(g), prompting me to finally check ICE’s 287(g) website.

What I found was shocking.

In the span of about two months, the Trump administration radically expanded the 287(g) program beyond anything I have seen in the past fifteen years of close study of this precise policy. The 287(g) webpage and table of active 287(g) signatories, which hadn’t changed substantially since at least 2008, has been completely redesigned. ICE provides templated requests that local agencies can use to request a 287(g) agreement; it’s as simple as plugging in the name of the agency. ICE now provides the list of active 287(g) agreements in an Excel spreadsheet (more convenient for me, so thanks, I guess), and ICE also now provides a list of pending 287(g) agreements. ICE produced new marketing materials for the program.

And there’s more. Much more.

First, the number of signed 287(g) agreements has exploded. As of April 11, there are 444 active 287(g) MOAs. That’s over three times the maximum number of MOAs that have ever existed. And that doesn’t count the additional 75 MOAs currently pending, which means we will soon see well over 500 signed 287(g) agreements. When Mario and I initially did our web scraping of the data, we thought 150 287(g) MOAs was a lot, especially compared to the upwards of 80 or so that previously peaked back in 2010.

In fact, ICE signed more new 287(g) agreements in a single week of the Trump administration so far than had ever existed at one time up to that point. During the week of February 24, 2025, ICE signed 140 new MOAs. In a single week.¹⁴

To see 500—and growing—is nothing short of astonishing. If it continues, which I’m sure it will, it will completely remake the landscape of interior immigration enforcement across the entire country.¹⁵ For now, new 287(g) agreements are heavily concentrated in Florida, Texas, North Carolina, and Georgia, but many other non-traditional states are signing on, too.

A searchable and downloadable list of all currently active 287(g) agreements is available at the end of this post.

Second, the explosion of 287(g) agreements is being driven in no small part by the revival of the most criticized and concerning type of agreement: the task force model.

As of today, ICE has signed 178 task force model 287(g) agreements—far above the previous maximum of around 40 back in 2010. This is on par with 180 active Warrant Service Officer agreements and more than the 86 active jail enforcement agreements.

Remember: the task force model effectively turns the cops you see in your community into immigration enforcement officers who can, in ICE’s own marketing materials, “carry out immigration enforcement activities ... during routine police enforcement duties.”¹⁶

To date, across all of these new MOAs, ICE has deputized 625 officers specifically under the task force model.

Although we still need to understand how this authority is being used (or will be used) in practice, at least from a conceptual perspective, this means that living in 287(g) county is equivalent to the kind of policing Border Patrol conducts in communities near the U.S.-Mexico border.

Third, the types of agencies signing on to 287(g) agreements have diversified wildly. In the past, 287(g) was primarily active at the county sheriff level and, to a lesser extent, some police departments and state prisons. Now new 287(g) agreements are being signed with agencies such as the following, and I’m only including those with task force models specifically:

• Florida Department of Financial Services

• Florida Fish and Wildlife Conservation Commission

• Kansas Bureau of Investigation

• Mississippi Attorney General's Office

• Montana Department of Justice

• Oklahoma Bureau of Investigation

• Oklahoma Bureau of Narcotics

• Texas Office of the Attorney General

• Florida National Guard

• Florida State Guard

We have never seen these types of state-level agencies get involved in 287(g) before; 287(g)s were always limited to front-line enforcement agencies or correctional institutions.

There is more to say about 287(g) and the data, and I will try to elaborate on this program in the coming days and weeks. (Please subscribe to keep up with that work.)

But we need to pull back for a minute to talk about what this means.

The staggering expansion of 287(g), and specifically the revival of the aggressive task force model, means that the Trump administration is building an army of deportation officers out of state and local law enforcement agencies at a scale that we have never seen before. Full stop.

I don’t mean that polemically or hyperbolically. I don’t even think that Thomas Homan would disagree with that framing. I think it is something that he and the administration is proud of—even though they have simultaneously been quiet about it. Bypassing constitutional restrictions on state enforcement and enrolling every law enforcement officer in the country into Trump’s mass deportation agenda is the goal. They are on their way to doing it. And no one is talking about it.

There is much left for us to understand. My previous research with Mat Coleman showed that the implementation of 287(g) can look very different in different jurisdictions. I also do not know all of the reasoning that local officials have for signing on to 287(g), nor do I understand how local policing practices are changing as a result of 287(g)—if at all.

That’s where you come in.

If you live in one of these states, counties, or cities, I would like to know your experience and perspective. Feel free to email, DM, or comment below. Immigration lawyers: that includes you.

If you are a reporter in a jurisdiction that has recently signed a 287(g) agreement, you have a unique opportunity to talk to your local law enforcement agency and elected officials about this program. The topic is massively under-reported right now. If you publish something, please share it with me so I can learn from your work.

Update 04/20/2025: Learn more about sheriffs from the following upcoming event.

Support Public Scholarship

This newsletter is only possible because of your support. If you believe in keeping this work free and open to the public, consider becoming a paid subscriber. You can read more about the mission and focus of this newsletter and learn why, after three years, I finally decided to offer a paid option. If you already support this newsletter financially, thank you.

Table of Active 287(g) Agreements

Find a searchable and downloadable table of 287(g) currently active agreements below.

1. The Illegal Immigration Reform and Immigrant Responsibility Act (IIRAIRA) was a deeply bipartisan piece of legislation signed into law by Democratic President Bill Clinton that dramatically altered the landscape of immigration enforcement in the United States. I highly recommend Dara Lind’s Vox article on IIRAIRA, which, like Dara, hasn’t aged a day since 2016. Pro tip for those who want to pass as an immigration insider: IIRAIRA is pronounced “EYE-ruh-EYE-ruh” and that is how everyone refers to it; as far as I can tell, no one who actually works in immigration research and policy remembers what the abbreviation is or what it stands for without looking it up—including myself.

2. The full text of this key part of 287(g) reads: “the Attorney General [now Department of Homeland Security] may enter into a written agreement with a State, or any political subdivision of a State, pursuant to which an officer or employee of the State or subdivision, who is determined by the Attorney General to be qualified to perform a function of an immigration officer in relation to the investigation, apprehension, or detention of aliens in the United States (including the transportation of such aliens across State lines to detention centers), may carry out such function at the expense of the State or political subdivision and to the extent consistent with State and local law.”

3. ICE used to refer to a 287(g) agreement as a memorandum of understanding (MOU) and currently uses the term memorandum of agreement (MOA). I use MOA and 287(g) agreement interchangeably.

4. This period was characterized by considerable experimentation in restrictionist policy at the state and local level. Hazleton, Pennsylvania passed local immigration policies in 2006 and Arizona passed SB 1070 in 2010—for just two examples.

5. The key word here is “relative.” New international migration of Latinos, as well as new internal migration of Latinos, gave rise to new migrant destinations in places like Raleigh, which had the second busiest Mexican consulate in the country when I was there in 2009, and Atlanta, which relied largely on migrant labor to build for the Olympic Games in 1996. This growth didn’t come close in absolute numbers to traditional destinations like Los Angeles and New York City. A theory emerged around that time that proposed that when it comes to explaining anti-immigrant backlash, a sudden growth from, say, 1 percent to 5 percent Latino/Hispanic generates a larger negative response than going from 40 percent to 45 percent, on account of the latter having undergone a type of normalization of ethnic diversity process that less diverse regions had not yet undergone.

6. I am ignoring, for the moment, the relationship between 287(g) and the Secure Communities program, which also emerged around this time. Secure Communities (and its subsequent iterations) was primarily a biometric information sharing program that did not delegate any enforcement authorities to local agencies, but which certainly did dramatically expand ICE’s ability to identify, detain, and deport immigrants in the country unlawfully.

7. I am tying my fingers in knots to avoid diverting this post toward the fascinating history of the office of the sheriff that runs from the “shire-reeves” of the British dark ages, through the Magna Carta, through the history of British colonial rule that gave us (and other colonial territories) the county and the sheriff (and, relatedly, the elected office of the coroner, by the way), to the role of southern sheriffs in the post-reconstruction and civil rights periods. 287(g) wouldn’t be possible without this fascinating historical and political geography.

8. I will shamelessly plug two of my articles that attempted to capture these developments, one on sanctuary church activism and another, with the amazing Angela Stuesse, on anti-deportation campaigns (academic article + Substack summary).

9. If you’re following along so far, you might recognize that one of the deep ironies of both the adoption and criticism of sanctuary city policies is that cities do not run the jails: counties do. So while a sanctuary city policy might send a positive message for immigrants and discourage immigration-based policing, the city has no formal control over the sheriff or the jail. That responsibility is typically left to county commissioners, and although it may have happened, I am not aware of any county commissioners adopting sanctuary policies. States, on the other hand—Maryland and Illinois come to mind—have tried to adopt state policies to restrict local participation in 287(g) agreements.

10. The EO Enhancing Public Safety in the Interior of the United States.

11. Maria Ramirez Uribe wrote an excellent profile of Homan for Poynter.

12. Although this is true, the combination of not doing on-the-ground research in Florida related to the WSO model and the effects of the pandemic make it hard for me to say with certainty what the overall effect was of this period of Warrant Service Officer agreements and of jail enforcement agreements, for that matter.

13. Huge thanks to colleagues at the crimmigration workshop in Eugene this weekend for letting me present these urgent findings, including especially Joel Sati, and also Juliet Stumpf, Hiroshi Motomura, Pilar Margarita Hernández Escontrías, Evelyn Rangel-Medina, Kit Johnson, Stacy Taeuber, Matthew Boaz, and Sabrina Balgamwalla. Also huge thanks to Mario Marset, Iona Volynets, and Adam Sawyer for their previous contributions to this research.

14. The first draft of this post mistakenly said that 180 MOAs had been signed on a single day, February 26, 2025. This was incorrect. About ninety MOAs were signed that day.

15. An important note here: because there are multiple types of 287(g) agreements, many agencies have more than one MOA, so 500 MOAs amounts to less than 500 agencies. I am not going to get into those details in this post.

16. The direct quote is from ICE’s new marketing materials for the Task Force Model.