Cooperation with ICE doesn’t stop community arrests: New data on the results of state strategies

The Trump administration has demanded local and state collusion on immigration enforcement, and promised to pull back on community arrests in return. But our analysis of updated ICE arrest data shows that collaboration has only expanded ICE arrests, and that the mass deportation project falters in places that resist.

by Jacob Kang-Brown and Brian Nam-Sonenstein, October 6, 2026

Table of Contents

ICE detainers turn local law enforcement into volunteer jailers

ICE arrests of all kinds have grown rapidly

Resistance has blunted surges in arrests

The federal criminalization loophole

Conclusion

Data sources

Appendix tables

The Trump admini­stration has spent its second term pressuring state and local officials to collaborate with the federal government’s mass deportation campaign. The reason why is simple: the administration cannot hope to get close to their arrest and deportation goals without the help of local law enforcement. Following the killings of Alex Pretti and Renee Good by ICE agents in Minnesota in early 2026, the administration sent out so-called “border czar” Tom Homan to make a deal: ICE agents would redeploy elsewhere if governments gave ICE access to people held in jails or deputized police to serve as federal agents.

However, our analysis of data from the Deportation Data Project shows this was a false bargain: ICE has spent 2026 arresting people en masse at their homes, in their neighborhoods, at their jobs, and elsewhere in their communities — regardless of whether their states collaborated with or resisted Trump’s mass deportation agenda. But we also find that state policy has nonetheless made a measurable impact on the scale and type of arrests ICE has been able to make, and states that have refused to comply have made headway in blunting the surge in arrests. This is especially true when it comes to arrests from “detainers,” which ICE uses to ask local law enforcement to voluntarily assist in handing over someone already in jail instead of releasing them. Detainer-related arrests require less effort and fewer resources than ICE raids and occupations, and have been a crucial tool for ensnaring criminalized people in the administration’s project. We found that states that have refused detainers have seen fewer arrests overall. In other words, the importance of detainer arrests to ICE — and their vulnerability to state laws — makes blocking this avenue a strategic opportunity for state governments to protect people from mass deportation.

Troublingly, our analysis also shows ICE continues to use the “federal criminalization loophole” to circumvent laws meant to prevent police from participating in immigration enforcement, expanding the number of people detained on federal immigration charges by 25% since the end of the Biden administration. To close this loophole, states and localities will need to go further and limit collaboration with the U.S. Marshals — not just ICE. ¹

In this briefing, we update our recent analyses of ICE arrest data, examining arrests through August 2026 and revisiting the impact different state approaches are having on immigration enforcement. Ultimately, we conclude that states can better fortify against federal immigration enforcement by banning collaboration with the U.S. Marshals, stopping police using biometric and criminal records checks from sending arrestee location data to ICE, and reducing police contact with the public. To supplement this briefing, we are publishing the underlying data that highlights ICE detainer-related arrests by state, as well as data on US Marshals detention by offense in appendix tables.

ICE is using detainers to turn local law enforcement into its volunteer jailers

Shocking images of ICE agents snatching people from cars and workplaces have rightfully fueled national outrage, but a less-visible, similarly devastating offensive has taken place behind jail walls. While the Trump administration has insisted it’s focused on targeting “the worst of the worst,” in reality most people in jails facing ICE detainers are being held pretrial and have not been convicted of a crime; most who have been convicted do not have serious criminal records. Nonetheless, our analysis shows that local jails play an essential role in boosting ICE arrest numbers through their voluntary compliance with these detainer requests.

The average number of daily detainer-related arrests has more than doubled since the last presidential administration, growing by 144% from President Biden’s last year in office in 2024 to Trump’s second year in office in 2026. ⁸ Between the first and second years of Trump’s current term (covering years 2025 and 2026), the average number of people arrested via detainer on a given day grew by another 16%, with detainers resulting in 376 people arrested on an average day in the first year, and 435 people arrested on average each day in the second year.

From the Trump administration’s perspective, detainers are an important tool for rapidly ensnaring people in the deportation machinery. Detainers drastically reduce the amount of work ICE has to do to arrest people — down to essentially making a few phone calls or emails — compared to deploying agents to stalk them in the community. Detainers make identifying and locating arrest targets relatively simple: ICE analyzes data that local law enforcement routinely and voluntarily transmit to the FBI (such as names and fingerprints) to check for outstanding warrants and criminal histories, looking for a match in its own databases and issuing a detainer if it finds one. ICE could not issue detainers without this information. Finally, even though ICE has a multibillion-dollar budget rivalling that of most nations’ militaries, they still lack the nationwide network of detention centers and jailers necessary to unleash their campaign of terror on immigrant communities on their own. Detainers bridge this gap, enabling ICE to snatch up immigrants who otherwise would be beyond their reach — all without having to pay local law enforcement extra money to do their work for them.

It’s reasonable for elected officials and community members to wonder why local law enforcement would take on the uncompensated, legally dubious work of serving as ICE’s volunteer jailers. Since unauthorized immigration does not cause more crime and increased immigration enforcement doesn’t reduce crime, this volunteer police work has no public safety benefits. But there may be other ideological or financial motivations at work. Executing detainers and locking people up until ICE gets there provides an on-the-job way for police to support the Trump administration. ⁹ These colluding officers may make low-level or even bogus arrests of noncitizens, and in this way bolster federal efforts to criminalize immigrants en masse. After an arrest by local police, ICE can more easily pressure immigrants to waive claims in immigration court and accept a fast track removal from the country. And while detainers themselves don’t cost anything, there are other substantial funding opportunities available for collaborators. ¹⁰

These kinds of low-level and bogus arrests are a longstanding feature of police work that have plagued poor and marginalized people in the U.S. for decades. Most people are jailed for public order, property, or drug charges – not “violent” charges — and many more are jailed for traffic violations or missing a court date. Even though the charges are relatively minor, their consequences can be devastating and destabilizing for many people regardless of their legal status. In the context of immigration enforcement, large numbers of people now face the additional, brutal harms of deportation for arrests stemming from truly minor violations.

ICE arrests of all kinds have grown rapidly since 2024

Now that we are well into the second year of the Trump administration, we have a clearer, fuller picture of how the mass deportation campaign has unfolded. As discussed in the previous section, arrests on detainer have risen dramatically since 2024, Biden’s last year in office. But nationwide, arrests in the community (such as in workplaces, homes, and on the street) have grown the most, followed by ICE arrests from a law enforcement setting (such as outside of a jail or at a courthouse) without a detainer. ¹¹ Overall, the data show two major escalations in ICE arrests.

The first escalation came as Trump took office and surged in the summer and fall of 2025. In the community, the average number of daily ICE arrests exploded, going from 82 per day on average at the end of 2024 to an astounding 458 arrests per day on average by mid-2026 — more than five and a half times higher. Meanwhile, ICE arrests from a law enforcement location (without a detainer) grew 455%, from 53 per day in 2024 to 294 per day in 2026. (For full details, see the table below.)

The second major escalation in arrests came in early 2026. Overall, average daily arrests grew by more than 28% between the first two years of Trump’s second term: 926 people were arrested by ICE each day on average in 2025, compared to an average of 1,187 per day in 2026. Drilling into the types or contexts of these arrests, we found that:

• Arrests in the community rose by 17%: an average of 390 arrests in the community per day in the first year of Trump’s administration grew to 458 per day in the second year.
• Law enforcement arrests without a detainer leapt by 84%: an average of 160 people a day were arrested this way in the first year, compared to an average of 290 per day in the second year. These arrests could be via informal collaboration that tips off ICE agents or via formal collaboration and deputization programs like the 287(g) agreement. ¹²

Table 1. National daily average of ICE arrests by type of law enforcement involvement.

ICE Arrest Type	Biden year 4	Trump year 1	Trump year 2	Percent change
	(2024)	(2025)	(Jan. to Aug. 2026)	2024-2026	2025-2026
Total ICE arrests	313	926	1,187	279%	28%
ICE arrests via law enforcement honoring detainer	178	376	435	144%	16%
ICE arrests at a law enforcement location	53	160	294	455%	84%
Community ICE arrests	82	390	458	459%	17%

Community and governmental resistance have blunted surges in arrests

While all types of arrests have surged in the last two years, state and local laws appear to account for key differences in how those arrests have played out on the ground. Following up on the analysis in our last briefing, we continue to see clear, persistent differences in arrest levels between (1) states with truly protective policies, (2) those that have tried to balance protections with enforcement, and (3) those that have gone all-in on collaborating with federal immigration enforcement. Appendix Table 1 contains the number of each type of arrest between 2024-2026 broken out by state, while Appendix Table 2 provides arrest rates.

States resisting ICE collaboration

While substantive protections have not fully stopped ICE from arresting people in the community, they have kept arrest levels of all types relatively low — especially when it comes to arrests involving law enforcement, with or without a detainer request. For example, in Oregon, only a small number of ICE arrests involve law enforcement, and the rare detainer arrest is usually connected to the U.S. Marshals or Federal Bureau of Prison facilities. ¹³ Oregon and Illinois have the strongest state policies to protect immigrants. ¹⁴

States collaborating with ICE

Unsurprisingly, some of the highest overall arrest numbers appear in states with laws that mandate collaboration and 287(g) deputization to assist in transport or transfer to ICE. States like Georgia and Florida also have mandatory immigration status checks at arrest, and rules for pretrial detention for unauthorized immigrants facing certain state charges so as to avoid release on bail before ICE can arrive. These policies have had their intended impact: the data show high levels of detainer-related ICE arrests, as well as arrests at law enforcement locations that are not connected to a detainer. Importantly, these policies have not prevented ICE arrests in the community, which occur at rates that are double those seen in Illinois and Oregon. ¹⁵

States trying to compromise between protection and collaboration

Some states like North Carolina require some law enforcement collaboration with ICE — such as immigration status queries at arrest and complying with ICE detainers — but do not go to the extremes of Florida or Georgia. Others like New Jersey make wishy-washy compromises when purporting to protect immigrants, instituting arrest charge-based carveouts (in other words, if police file paperwork a certain way, they can work with ICE and ensure transfer to ICE custody). Despite having a less enthusiastic policy around collaboration, New Jersey has arrest levels nearly comparable to Georgia’s. Notably, community arrest levels in both North Carolina and New Jersey are comparable to those of states that have gone all-in on collaboration, meaning their compromises utterly failed to keep ICE agents from stalking their neighborhoods.

ICE is still circumventing limits on collaboration via the “federal criminalization loophole”

The Trump administration continues to dodge state limits on law enforcement collaboration by using the “federal criminalization loophole.” Under this strategy, ICE and other federal agencies refer people for federal prosecution for immigration-related “crimes,” ¹⁶ moving them into the custody of the U.S. Marshals and thus enabling them to use contracted local jails even if they’re in sanctuary cities, counties, and states. Once those people are finished with their sentence or the prosecution is dropped, they are brought back into ICE custody. While no state law can prevent federal agencies from working together, states can prevent federal agencies from using local jails in their chain of custody.

Unlike detainers, local governments are paid a per diem amount to jail people for the federal criminal courts. The Marshals rely heavily on this existing infrastructure: only a small share of people in their custody are held in federal facilities, with the rest incarcerated or detained in the Marshals’ detention network composed of local jails, state prisons, and private prison facilities.

Our analysis shows that, among people held for the Marshals in May 2026, the number of people detained for federal immigration charges is up 25% compared to fiscal year 2024, while detentions for all other reasons (drugs, weapons, property or violent crime, supervision violations, etc.) have fallen by 11%. After an initial surge at the start of Trump’s second term, the percentage of people held for immigration charges fell temporarily but has been rebounding in 2026.

Fortunately, states and localities can close this loophole by not just banning collaboration with ICE, but with other federal law enforcement agencies that have been deputized into the mass deportation agenda as well. This should be an urgent priority for state and local governments: new data from the U.S. Marshals show that directives to U.S. attorneys that prioritize criminalizing immigrants over all other issues are having a real impact on who is detained and incarcerated. There are now almost 20,000 people in federal custody who are detained primarily due to federal criminal charges for immigration — around 13,000 by the Marshals and 7,000 at the Bureau of Prisons — a nearly 12% increase since 2024. ¹⁷ (These sums do not include people with unauthorized immigration status who are prosecuted for other crimes – like fraud related to passports).

Conclusion

While even the most robust protections governments have offered to immigrants have not halted ICE arrests entirely, they have made a meaningful difference in the scale of arrests overall. The relative success of these strategies points to the importance of preventing law enforcement from joining forces with the federal government to arrest, criminalize, and detain immigrants. Simply put, ICE will struggle to execute the mass deportation campaign without the personnel and temporary detention capacity provided by state and local law enforcement. And while ICE is racing to hire more people and bring more detention centers online, they have struggled to deliver on their plans. The more states can strengthen their resistance to this co-optation of state and local law enforcement, the longer it will take ICE to catch up.

The good news is that there’s room to strengthen existing tactics aimed at preventing collaboration, and there are other ideas that have yet to be tested.

Strengthening prohibitions on law enforcement collaboration

Expanding sanctuary laws to cover federal immigration enforcement in general — and not just collaboration with ICE — is one promising avenue for lawmakers to pursue. Oregon’s sanctuary law expanded the definition of immigration enforcement in 2021 to include federal criminal prosecution “for offenses related to federal laws regarding immigration status.” ¹⁸ Advocates and litigators have sued Oregon counties with U.S. Marshals contracts for detention space in their jails, seeking to close the federal criminalization loophole. ¹⁹ This appears to be making an impact: In Oregon between February and June 2026 (the four months following the lawsuits), the number of people charged by federal prosecutors with reentering the U.S. after deportation — the most commonly charged federal criminal offense related to immigration — decreased by a third when compared to the same period in the prior year. ²⁰ During this time, neighboring Washington state and the Northern District of California saw upticks in those prosecutions. ²¹ Although state law cannot stop agents from taking someone already in federal custody to a different federal court in another state, there are limits on prosecuting someone in a place other than where they were initially arrested. ²² So far, limiting federal access to local jails appears to work.

Oregon’s approach is more effective than those seen in other states. Take, for example, New York’s recent law, which prohibits counties from making agreements to hold people for ICE: the “Local Cops, Local Crimes Act” included in the 2026 budget deal specifically authorizes detention contracts for people with federal criminal charges like illegal reentry. While some protections are arguably better than none, approaches like this in New York leave room for federal agents to exploit the criminalization loophole, rendering them far less effective than they could be.

Limiting data sharing with the federal government

The common police practice of sharing location data with the federal government as part of criminal history checks and facility transfer processes represents an underappreciated weakness in sanctuary laws; it is a form of collaboration baked into everyday police work that persists even in states that have the strongest protections. But states have the ability to regulate the federal government’s access to these records, and can govern access for non-criminal history purposes. Many states already have regulations that limit public access or statistical research projects, and could potentially add restrictions on their use in enforcing federal immigration law. Especially in more progressive states, advocates could seek to further regulate the use of identity check procedures when people are transferred from one agency to another, such as from the local police lock up to the county jail. This is important because without routine, voluntary, and extraneous data sharing by law enforcement, ICE could not issue detainer requests. There is currently no federal requirement that state and local police fingerprint everyone arrested and contribute their biometric information to the FBI database. Though we do not know of any states that have pursued this strategy to date, this kind of excessive data sharing raises privacy concerns, and limiting the federal government’s access to state and local records would likely greatly inhibit ICE’s ability to identify targets for arrest.

Reducing police contact

As we have discussed, ICE first learns of many eventual arrest targets through the policing of nonserious, low-level offenses. This is a longstanding issue that goes beyond immigration and is responsible for millions of jail bookings each year, most often involving poor people and people of color. Both for the sake of blunting ICE arrests and for the general quality of life of people in their jurisdictions, governments should more seriously consider decriminalization of many of these offenses and reducing police contact where possible.

As the Trump Administration continues to arrest and detain immigrants across the U.S. under the guise of going after “criminals,” many states and local governments have sought to protect their communities by prohibiting law enforcement collusion with ICE. Others have acted to prevent ICE detention in their communities. Many states recognize that conflating immigration and criminal legal system policy is harmful.

But more can be done to reduce police contact overall and send fewer people into the system. Armed with stronger protections, states might want to go on the offensive against ICE’s use of detainers by changing procedures for state criminal history records checks. They could also block the federal criminalization loophole by canceling contracts between their jails and the U.S. Marshals. Together, state and local governments can help create a bright line between the civil immigration system and the criminal legal system.

Data sources and methodology

Data sources

ICE arrest, detention, and detainer request data were obtained via Freedom of Information Act request and processed by the Deportation Data Project, which were then analyzed by the Prison Policy Initiative.

Data on the average number of people detained while facing criminal charges in federal courts or otherwise incarcerated by the U.S. Marshals Service come from monthly reports provided by the Department of Justice to the U.S. Congress Committee on Appropriations. A full data table and the source PDF files are available here.

Data on people facing federal criminal prosecutions for immigration-related matters (such as violations of 8 USC S 1326) are drawn from the Department of Justice Offices of the United States Attorneys’ Prosecuting Immigration Crimes Report, which covers cases through June 2026. These data include all defendants, regardless of whether the immigration charge is the most serious charge or not.

Population data used for calculating rates come from the U.S. Census Bureau’s Annual Estimates of the Resident Population for the United States, Regions, States, District of Columbia and Puerto Rico: April 1, 2020 to July 1, 2025 (NST-EST2025-POP).

Data corrections and processing

The Deportation Data Project has secured an ICE arrest database table through August 5, 2026. Building upon this important work, we calculated arrest rates, grouped apprehension locations by broader categories, filled in missing apprehension state data where possible, and processed data to remove any obvious duplicate records. The ICE arrest data tables do not include all immigration-related arrests: for example, they do not include Customs and Border Patrol arrests, which are especially common in some states along the U.S.-Mexico border or in places like Vermont or upstate New York. It also appears that in relatively rare cases, data entry gaps mean that people that were held on a detainer, arrested by ICE, and booked into detention may not always have an arrest record.

The Deportation Data Project identifies some records that appear to be duplicated in the arrest table, so we removed extra rows that involve a person being booked into ICE custody more than once in a single day. ICE arrest records are not always complete: in the period covered by our analysis, 7.3% were missing state information, enough to influence the analysis. ²³ We filled in 89% of missing state locations using a combination of the ICE office area responsible and the information provided about specific arrest locations, such as “apprehension city” or “event landmark” (a mix of information, but often includes names of facilities), as well as any detention stay records. ²⁴ Sometimes ICE arrests may be processed in a state that is different than where officers actually encountered the person, so these estimates may incorrectly assign ambiguous cases in regions like Northern Virginia and Washington D.C., or the Kansas City metropolitan area.

In order to process the ICE detainer request data for analysis, we removed duplicate records and included only the most recent detainer issued for each person. If there was more than one detainer recorded for a person on a single day, we prioritized the detainer that was associated with an ICE arrest, recognizable by the record having an “apprehension date.” (The alternative, the duplicate” filter provided in the Deportation Data Project’s processed version of the ICE detainers file, uses the sequence of rows in the original FOIA data table to address duplicate records; we determined that this could lead to discarding detainers that match to an arrest, and therefore prefer our method.) This helped address issues like multiple detainers for the same person — for example, one at initial booking at a municipal police department followed by a second detainer issued when the person was transferred to the county jail.

We also excluded records that did not have a valid personal identification number. This might lead to an undercount of records if someone is held on a detainer but has never had prior contact with immigration authorities, and so does not have a personal identification number used by the Department of Homeland Security. However, it solved other problems caused by spurious and duplicate records, and allowed matching records from the detainers dataset into detention and arrest data tables. It seemed clear from careful review of the dataset that some records are erroneous, and appear to be created as part of the process of trying to secure a match between booking formation and DHS records. ²⁵

In order to determine whether ICE arrests were related to detainer requests or involved local jails, state prisons, police, sheriffs or another law enforcement agency in some other way, we used a variety of data fields across the ICE detainer and the ICE arrest tables from the Deportation Data Project to make the determination. The category, “Workplaces, homes, courts, and others” is a residual, catch-all category defined in relation to our other two categories, (1) detainer-related arrests and (2) other criminal legal-related arrests.

We defined ICE detainer requests that led to an ICE arrest as those where the field “Detainer Lift Reason” was coded as “Booked into Detention.” After careful inspection, we found records for thousands of people that were in the detainer data and coded as booked into detention but did not match with any records in the ICE arrest data. Nonetheless, they matched with either stays in detention or other data that indicated the people were removed or deported from the United States. ²⁶ Inspection of the records indicated that they were clustered in time, and appeared to be connected to staffing gaps in data entry. For example, amidst a consistent surge of detainer activity in South Florida in early July 2026, there were a number of records that should have been in the ICE arrest file that were not on the day before July 4. Thus, we assume that these are legitimate records and their missingness from the ICE arrest table is accidental, likely the result of staffing issues.

We defined ICE arrests that involve law enforcement in some other way by cross-classifying the apprehension method and event landmark fields in the arrest table. ICE data entry practices do not appear consistent across field offices and have changed over time, ²⁷ and this analysis is the result of a direct, manual review of many records. We accounted for misspellings in the “event landmark field” and for the incomplete nature of coding in the “apprehension method” field. This ensured we did not undercount the connections between local law enforcement and ICE. We classified arrest “event landmarks” into the criminal legal-related category if they mentioned jails, prisons, sheriff departments, probation or parole agencies, 287(g), or similar terms, using an iterative process that checked and accounted for all variant spellings. We also classified arrest “apprehension methods” as criminal legal-related if they involved 287(g) or the so-called Criminal Alien Program, custodial arrests, the Law Enforcement Agency Response Unit, or probation or parole.

28-day rolling average: This metric helps smooth out the daily data variation into trends, and was calculated using a sum of all arrests over four weeks divided by 28 to get the average daily arrest rate in the period. For example, July 30’s number is the average number of arrests for July 3-July 30.