The record on immigration enforcement conduct, January 2025–July 2026

Retrospective — written July 2026, covering January 2025 to publication. Assembled from the sources cited below; not contemporaneous coverage.

What this thread contains

A caveat first, because it governs how far this entry can go. Much of the research payload assembled for this thread is not reporting on immigration operations at all. It is search-tool commentary noting that ICE has not adopted or referenced any "democratic backsliding" framework, which describes what the searches returned rather than anything ICE did. That material is set aside here. What remains is a genuine but partial dated record: several primary directives, a handful of documented enforcement tactics, and secondary characterizations from legal and advocacy sources. Where the payload is thin (notably on granular, dated enforcement statistics for interior operations), this entry says so rather than filling the gap.

On January 20, 2025, per the White House order "Protecting the American People Against Invasion," ICE was directed to prioritize enforcement of final orders of removal, redirect its Homeland Security Investigations division toward Immigration and Nationality Act violations, and reestablish the VOICE office for victims of crimes by removable aliens. On January 21, 2025, per the Department of Homeland Security, Acting Secretary Benjamine Huffman issued directives ending what DHS termed humanitarian-parole abuses and rescinding Biden-era guidelines that had limited ICE and CBP enforcement in "sensitive" areas such as schools and churches.

In the same early window, ICE was directed, per the American Immigration Council's May 2025 FOIA request, to detain noncitizens "to the maximum extent authorized by law" and to realign parole policies. The Council's central documented complaint is procedural: ICE did not publish formal policies implementing these orders, which the Council characterizes as a failure of the agency's FOIA obligation and an obstacle to public assessment. Separately, per Just Security's expert survey, an internal memo from the Acting Director of ICE moved the agency toward using administrative warrants to enter homes for immigration arrests without judicial authorization.

In March 2025, per NAFSA, DHS carried out a reduction in force affecting more than 100 employees across three oversight offices, including the CIS Ombudsman's Office. Across 2025, per the Texas Law Review, changes reached the immigration-court system directly: the administration dismissed immigration trial judges and 13 Board of Immigration Appeals members, reducing the Board from 28 to 15; EOIR withdrew the policy that had prevented arrests at immigration courthouses, leading to arrests that disrupted hearings; and a 2025 EOIR regulation removed the requirement that temporary immigration judges hold ten years of immigration-law experience. The same source records that by the end of 2025 the Board's backlog reached roughly 220,000 cases and describes the system as facing significant risk of political influence.

On the interior-enforcement side, per the New York State Bar Association, ICE intensified worksite enforcement: more I-9 audits, document requests, and compliance reviews, in some instances triggered by data discrepancies shared across agencies including the IRS. Per The Fulcrum, the administration justified these expansions under a unitary-executive theory of presidential discretion to allocate detention and removal funds, a theory critics tie to due-process and appropriations concerns. On June 6, 2025, per KTLA, protests against federal immigration enforcement began in Los Angeles after a series of workplace raids, with demonstrators clashing with local police and federal agents outside a downtown detention facility.

Later in the period the thread surfaced in the appropriations fight. Per the House Appropriations Committee, Republicans accused Senate Democrats of weakening authority over ICE and CBP; per KCRA, Democrats demanded limitations on ICE enforcement, judicial warrants, and bans on racial profiling, calling the administration's proposals inadequate. Issue One's "Unchecked Exec" report (roughly the 100-day mark, May 2025) listed "defy court orders in immigration cases" among ten areas of overreach but, as it and the payload both note, supplied no specific dated cases. This entry cannot close that gap from the record in hand.

The record's shape

The trajectory is one of steady, front-loaded expansion rather than escalation over time. The defining moves (the parole and sensitive-locations rescissions, the maximum-extent detention directive, the courthouse-arrest reversal, the oversight and adjudicative-personnel changes) cluster in the first quarter of 2025 and then persist. The payload does not show the conduct intensifying across 2025–26 so much as being established early and left in place; the later entries (the June raids, the appropriations fight, the year-end backlog figure) are consequences of the early architecture, not new escalations in legality or scale.

On base rates, the thread splits into categories that resolve differently, so each is anchored separately per the base-rate test. The enforcement-volume and worksite categories have named same-category comparators that the current record matches rather than exceeds: Obama's roughly 3.08 million formal removals (FY2009–2016) set a bipartisan high-water mark for removal volume, and George W. Bush's 2008 Postville raid (389 arrested, ethnicity-based detention, fast-track prosecutions criticized on due-process grounds) is a direct conduct precedent for the June 2025 Los Angeles workplace raids. This is why the thread does not read as pattern: its spine is precedented in kind. The oversight-reduction category is adjacent-precedented (Reagan's 1981 removal of some fifteen inspectors general is the transition-turnover comparator), but the payload's reach into the adjudicative function (dismissing Board members, cutting judge-experience requirements) has no comparator supplied here, and the administrative-warrant-without-judicial-authorization practice has none either. Those are the sub-categories that keep the thread from reading as merely precedented.

The contrary and mitigating evidence carries real weight and is not confined to a caveat. Civil society operated as the framework's calibration expects: the American Immigration Council filed FOIA litigation and is pressing the non-publication of guidance through the courts; protest activity in Los Angeles proceeded; and Congress engaged directly, with one chamber's minority demanding statutory limits, judicial-warrant requirements, and racial-profiling bans as conditions in the appropriations fight. These are functioning checks, present in the record, and they are the strongest reason the reading is contested rather than pattern. Against them sits the FOIA non-publication itself, which cuts the other way. It narrows the very transparency those checks depend on.

The bundle check weakens rather than strengthens a pattern reading for this thread specifically. Only one indicator co-moved in the period the payload covers: violence, which sits at pattern on the strength of the January 6 clemency grants. That is an unrelated mechanism (clemency for completed political violence, not immigration enforcement), so its movement does not corroborate a selective-enforcement or civil-service pattern here. Principle 5 asks whether co-movement tightens the signal; in this case the co-moving indicator is orthogonal, and the two indicators this thread is scoped to both stand at contested, undisturbed by the assembled record.

Measured against the indicator definitions, the record does not clear the bars either indicator names for moving off contested. Selective-enforcement requires "a documented, data-supported pattern of enforcement disparity targeting a specific group or critics beyond historical baseline rates"; the payload documents aggressive tactics and due-process concerns but supplies no such dataset and no court finding of selectivity. Civil-service requires "a substantiated pattern of loyalty-test-based removals extending materially beyond normal appointee turnover"; the oversight and judicial dismissals are documented as to fact but not characterized in the payload as loyalty-tested rather than restructuring. Both bars name a specific kind of evidence, and that evidence is absent. That is why no standing-reading change is proposed. The honest reading is that eighteen months of this payload's record establishes an aggressive, early-consolidated enforcement posture with genuine due-process friction, resting on categories with real bipartisan precedent, and does not, on the evidence in hand, discriminate between a good-faith statutory-execution account and a selectivity account. That is a contested record, and this entry reflects the payload available to it, not a claim that no discriminating evidence exists anywhere.