ICE Arrests 75+ U-Visa Crime Victims During Operation Metro Surge, Reversing Decades of Protection Policy

The Minnesota Reformer identified at least 75 people detained during Operation Metro Surge who held U visa status — work permits and deportation protections granted specifically to immigrants who are crime victims cooperating with law enforcement. Victims arrested include survivors of shootings, stabbings, human trafficking, felony assault, and armed robbery. The reversal traces to a policy change: ten days after Trump’s second inauguration (2025-01-31), ICE’s acting director, Caleb Vitello, issued a memo — ICE Policy 11005.4 — overturning decades of agency guidance that had barred arresting immigrants with pending U-visa applications. See 2025-01-31–ice-vitello-memo-rescinds-u-visa-victim-arrest-protections for the primary document and what it specifically rescinded.

Counting method, stated explicitly: Minnesota Reformer’s 75-count is not an ICE-disclosed figure. It comes from the Reformer’s own review of more than 1,100 habeas corpus petitions — federal lawsuits challenging detention — filed in Minnesota between December 2025 and mid-March 2026. The Reformer explicitly flags the count as a likely undercount: many detained people lacked attorney access for long stretches, could not afford legal fees, or were transferred out of state before a Minnesota habeas petition could be filed on their behalf. This KB should cite “at least 75, Minnesota Reformer’s own count via habeas-petition review, likely an undercount” — not a bare “75” presented as comprehensive.

Independent corroboration exists at the individual-case level. 2026-08-03–ramirez-mancia-habeas-release-freeborn-county documents, via primary federal court docket rather than news narrative, one named U-visa petitioner (Kevin Ramirez Mancia) held four months in Freeborn County Jail after an unrelated DUI stop, before winning habeas release — independently confirming the pattern the Reformer’s petition review found at scale.

Litigation context: a nationwide class action, 2026-05-20–icwc-v-noem-preliminary-injunction-u-visa-arrest-protections (ICWC v. Noem, C.D. Cal., decided 2026-05-20, roughly three months before this arrest count was published), obtained a preliminary injunction against enforcement under the same Vitello memo and certified three nationwide plaintiff classes. Per Kansas City immigration attorney Rekha Sharma-Crawford, arrests of U/T-visa holders “slowed” after that ruling but did not stop; the government has continued to litigate the injunction on appeal.

This is a distinct finding from the operation’s previously documented 1,300 “collateral” arrests (2026-04-08–ice-data-reveals-1300-collateral-arrests-operation-metro-surge) — those were bystanders swept up incidentally, while the U-visa arrests target people the government had formally recognized as crime victims and law-enforcement cooperators. Operation Metro Surge has generated a sustained record of documented harm in this corpus, from its initial 3,000-arrest announcement (2026-01-19–ice-announces-3000-arrests-operation-metro-surge) through an agent’s criminal assault charge (2026-04-16–moriarty-charges-ice-agent-morgan-assault-minnesota). The 75-victim figure itself remains tier-2 sourced (Minnesota Reformer’s own count); the policy instrument enabling the arrests and one individual case within the count are now tier-1 sourced.

Scope limit on how victims were located — stated so no downstream piece overclaims. The sharpest available version of this story would be that ICE mined pending U-visa applications to build a target list. That is NOT established by the record here and must not be written. What the documentation supports is narrower and still serious: USCIS check-in appointments and deferred-action grant records placed people at known times and places in ICE’s path. Application-data mining and appointment-based location are different mechanisms with different evidentiary requirements; only the second is documented. Anyone drawing on this entry should carry the distinction.

How the count was produced. The “at least 75” figure is Minnesota Reformer’s, derived from its own review of 1,100+ habeas petitions filed in Minnesota — a method the Reformer itself describes as yielding a likely undercount. This corpus has not independently re-parsed that petition set (PACER cost), so the figure is reported as the Reformer’s finding on the Reformer’s method, not as our own count. A KB-independent number would require parsing the docket list directly.