How did ICE arrest categories (criminal vs civil) change between 2023 and 2025?
Executive summary
Between 2023 and 2025 ICE’s public arrest reporting and the underlying composition of who was being arrested shifted markedly: official categories and coding practices were changed or obscured by the agency while the share—and absolute number—of arrests of people without recorded criminal convictions rose sharply under new enforcement directives and quotas [1] [2] [3]. Independent datasets obtained via FOIA and analyzed by researchers show both a methodological reclassification of many jail-based arrests into a generic “custodial arrests” label and a real operational pivot toward community and non‑criminal arrests driven by policy changes in 2025 [1] [4] [5].
1. How ICE defined “criminal” vs. “civil” in 2023 and the baseline data
In the publicly available ICE statistics through 2023, the agency distinguished arrests by criminal-history flags—generally treating immigration violations as “civil” enforcement while highlighting arrests of individuals with criminal convictions as distinct categories—information that fed historical trends showing ERO typically arrested immigration violators with convictions for DUI, drug possession, assault and certain traffic offenses [6] [7]. Independent archives and research projects preserved those classifications and allowed longitudinal comparisons through 2023, enabling scholars to trace pathways from arrest to detainer to detention [8].
2. Coding changes in 2025 that obscured prior categories
Beginning in 2025 ICE altered how it reported the locus and program source of arrests: according to analyses by the Deportation Data Project and the Prison Policy Initiative, as of the last week of July 2025 ICE stopped reporting many arrests under the various local, state, and federal “Criminal Alien Programs” and instead lumped them into a generic “Custodial Arrests” category, a change that reduces transparency about whether detentions originated from jail-to-ICE transfers or other programs [1]. Multiple FOIA-derived datasets through October 2025 confirm ICE’s shifting coding practices and researchers warn that data corrections and new defaults complicate apples‑to‑apples comparisons between 2023 and 2025 [8] [9].
3. The operational shift: more at-large and non‑criminal arrests in 2025
Concurrently with the reporting changes, enforcement tactics shifted: researchers documented a move away from predominantly jail-based intake toward community or “at-large” arrests and an increase in arrests of people with no recorded criminal convictions—roughly one in three arrests in parts of 2025 according to analyses cited by advocacy groups and legal clinics—driven in part by internal ICE quotas introduced in May 2025 and by policy directives that loosened protections for “sensitive locations” such as courthouses and hospitals [2] [3] [10] [5]. Public interest reporting and legal analyses also registered record monthly detention and arrest volumes in mid‑2025, with June and later months seeing daily arrest averages climb into the high hundreds or thousands [11] [2].
4. Data-quality caveats and competing interpretations
Experts and watchdogs note that some of the apparent rise in “non‑criminal” arrests is amplified by shifts in ICE’s internal flags and by problems in location and classification metadata—meaning part of the trend reflects changed recordkeeping as much as operational priorities—and ICE’s own statistics and dashboards were reworked in 2025 in ways that make longitudinal comparisons fraught without careful reconciliation by researchers [1] [9] [12]. ICE’s public statistics still describe the categories they use, but independent FOIA releases and third‑party dashboards (Deportation Data Project, Vera, TRAC) are essential to unpack how much of the change is true substitution of non‑criminal for criminal targeting versus recoding of similar events [6] [8] [12] [13].
5. What this means for accountability and policymaking
The combined effect of coding changes plus a demonstrable operational expansion in community arrests and non‑criminal detentions between 2023 and 2025 has reduced the clarity of oversight: researchers say the move to generic “custodial” labels and the surge in arrests without criminal records make it harder to assess whether ICE is meeting any stated policy priorities or to evaluate local‑state cooperation’s role in deportation flows [1] [2] [4]. Alternative views exist—ICE’s public materials continue to assert categories and justifications for arrests in their statistics—but independent FOIA-backed datasets and reporting demonstrate that both classification and practice evolved substantially in 2025, and that robust comparison to 2023 requires careful use of those external datasets [6] [8] [4].