Civic Shield — A U.S. Immigration Policy, Assistance and Monitoring Project
Civic Shield exists to put the facts about U.S. immigration enforcement in one place, with every number traced to its source and no political side taken. It is built for immigrants who need to know their rights and where to turn, and for citizens who want to understand what is happening in their communities. Where the record is unclear or contested, this site says so.
The numbers and the story
U.S. immigration enforcement, January 2025 to today. Every figure on this page links to the page it came from. Where the government's number can't be checked, it says so. Last checked by a person on .
In one glance
- 65,765
- in ICE detention, Jul 11, 2026 — the record was 70,766 in late Jan 2026 TRAC
- 50,925
- ICE arrests in Aug 2026, the third monthly record in a row, per DHS ExecutiveGov
- 356,389
- removals in fiscal 2026 through mid-July, per ICE ABC News
- 70.6%
- of detained people have no criminal conviction TRAC
- 37
- deaths in ICE custody in fiscal 2026, the most on record Bloomberg Law
- 1,202
- 287(g) agreements with local police, Dec 2025, up from about 120 a year earlier Kocher
What the government reports but no one can check
Is it ramping up, flat, or falling?
Both. Detention roughly doubled over 2025, peaked at about 70,800 in January 2026, dropped to 60,300 by April as removals and the self-deportation program caught up with intake, and has climbed back toward the peak since. Arrests, by contrast, kept setting records through the summer of 2026: more people are being arrested, but a growing share are released, monitored, or stuck in a backlog rather than removed. Reuters reported in September that removals have held near 1,200 a day while arrests passed 1,600 a day.
The pivot from border to interior
In November 2024, two-thirds of people entering ICE detention had been arrested by Border Patrol at the border (13,572 of 21,130). By June 2026 it was under one in twelve (3,575 of 43,138): ICE's own interior arrests now drive detention. TRAC Jan 2025 · TRAC Quick Facts
The Marshall Project traced much of the 2026 surge to 287(g) partnerships: traffic stops and jail bookings by local police in participating states, rather than large visible operations.
Recent reporting, five and five
Five stories the administration and its supporters point to, five that critics point to, each linked to the article itself. Reviewed and refreshed weekly. Some outlets block direct links; those are linked through a member station or syndication partner carrying the same article.
Loading news…
Georgia
Loading state information…
Who is being detained
Read these two lines together. About seven in ten people in detention have never been convicted of a crime. About four or five in ten have neither a conviction nor a pending charge: their only violation is a civil immigration one. The rest have a charge pending, which is counted as "no conviction" by TRAC and as "criminal" in DHS statements. That definitional gap is why the two sides quote different numbers from the same data.
- 52 days
- average stay in detention, fiscal 2026 through July, up from 44 days in September 2025. Longer stays plus record intake is what drives the population up. Open Immigration (aggregator of ICE spreadsheets) · Migration Policy Institute
- 183,181
- people monitored outside detention (ankle monitor, phone app, check-ins), Jul 2026. TRAC
- 469
- facilities held people for ICE in Jul 2026; 208 were listed on ICE's website. Congress funded about 41,500 beds; the rest is paid for through emergency transfers. Vera
Policy changes and where they stand
Status as of early October 2026. "Enjoined" means a court has blocked it while the case continues. Click the source for the rule or ruling itself.
Immigrant visa pause for 75 countries
The State Department stopped issuing green-card visas at consulates to nationals of 75 countries, citing public-charge risk. People can still apply and interview, but no visa is issued. Work, student and tourist visas are not affected.
Affects: family- and employment-based immigrant visa applicants abroad from the listed countries, including Brazil, Egypt, Ghana, Iraq, Jamaica, Pakistan, Russia and Ukraine.
Mintz, Jan 29, 2026H-1B: $100,000 fee and wage-weighted lottery
The $100,000 payment on new H-1B workers (proclaimed Sept 2025, renewed Sept 2026) was vacated by federal courts in Massachusetts (June 2026) and California (Sept 30, 2026). Separately, since Feb 27, 2026 the H-1B lottery favors higher-wage jobs, and DHS has proposed a $103,265 fee through formal rulemaking.
Affects: employers sponsoring new H-1B workers; entry-level foreign graduates most.
Forbes, Oct 1, 2026 · BakerHostetlerEnd of "duration of status" for students
A DHS rule would cap F-1 and J-1 stays at four years, shorten the grace period from 60 to 30 days, and require paid extensions through USCIS. A federal court blocked it the day after it was to take effect.
Affects: roughly a million international students and exchange visitors, and their dependents.
Auburn University, Sept 21, 2026TPS terminations and end of CHNV parole
DHS has ended Temporary Protected Status for 11 countries covering about 1.04 million people. Haiti and Syria ended July 27, 2026; Honduras, Nepal and Nicaragua earlier; Ethiopia and Somalia remain blocked. Venezuela's 2021 designation is before the Supreme Court; its work permits expired Oct 2, 2026. The CHNV parole program (Cuba, Haiti, Nicaragua, Venezuela) ended after a May 2025 Supreme Court stay.
Affects: ~615,000 Venezuelans, ~335,000 Haitians, ~51,000 Hondurans with TPS; ~500,000 CHNV parolees.
Pew, Aug 12, 2026 · AILAExpedited removal nationwide and no-bond detention
Anyone in the interior who cannot prove two years of continuous presence can be deported without seeing an immigration judge; the D.C. Circuit let this take effect in July 2026. Separately, the BIA's Matter of Yajure Hurtado denies bond hearings to people who entered without inspection. Courts in California and the Third Circuit have rejected it; the Fifth and Eighth uphold it, so bond eligibility now depends on where you are held.
Affects: long-term residents without proof of two years' presence; detained people who entered without inspection.
AILA, Jul 2026 · Yajure Hurtado status, Aug 29, 2026Public charge rule rescinded; social-media vetting
DHS withdrew the 2022 public-charge rule, so officers may weigh any cash or non-cash benefit when deciding green-card applications. Visa applicants in most categories must set social media to public, and USCIS now collects social-media identifiers on nine forms including the I-485 and N-400.
Affects: green-card applicants and mixed-status families; most nonimmigrant visa applicants; citizenship applicants.
Federal Register, Jul 20, 2026 · Social media screeningTougher I-9 and E-Verify enforcement
ICE reclassified more than ten paperwork errors on Form I-9 from "technical" to "substantive," removed the ten-day chance to fix them, and set fines of $288 to $2,861 per form. Employers using remote document checks must prove they were active E-Verify participants.
Affects: every U.S. employer; workers whose employers are audited.
Morgan Lewis, Apr 14, 2026287(g): local police as immigration officers
Agreements that let local police and sheriffs question, detain and transfer people for ICE grew from about 120 in January 2025 to 1,202 by December 2025, most under the "task force" model that allows street-level enforcement. Georgia had 45 participating jurisdictions as of February 2026.
Affects: anyone stopped or booked by a participating agency, including at traffic stops.
Kocher, Dec 9, 2025 · ICE 287(g) pagePaid "self-deportation" (CBP Home / Project Homecoming)
People who register their departure through the CBP Home app get travel paid and a stipend, raised from $1,000 to $2,600 in January 2026. About 72,000 had used it by March 2026 and roughly 132,000 by mid-July; around half came from inside ICE detention. DHS says it costs about $2,500 per case versus about $18,000 for a formal removal. Lawyers warn that leaving can trigger multi-year re-entry bars that the program does not waive in every case.
Affects: undocumented people weighing whether to leave voluntarily.
AILA, Jan 2026 · Axios, Jul 28, 2026"Gold Card" green card for a $1 million donation
A nonrefundable $1 million payment to the Treasury counts as evidence of extraordinary ability for an EB-1 or EB-2 green card. Uptake is small: by May 2026, 165 people had paid, 59 had filed, one was approved. No separate STEM or researcher fast-track policy could be verified; claims of one appear to be unfounded.
Affects: wealthy applicants only; no change for ordinary EB-1/EB-2 cases.
Documented, May 15, 2026Both sides of the argument
The strongest version of each case, with who is making it. Where the two sides use the same data, the dispute is usually about definitions.
The case for the enforcement surge
- Deterrence works at the border. DHS reports 16 consecutive months with no releases at the southwest border and apprehensions down about 94%, which it attributes to credible interior enforcement. Fox News
- The law is finally being applied. Supporters argue that final removal orders and unlawful presence are violations regardless of criminal history, and that prior administrations' discretion amounted to non-enforcement. The expedited-removal expansion has been upheld on appeal. AILA
- Public safety. ICE and DOJ highlight removals of people with gang affiliations and serious convictions; ICE says 720 of 1,226 people arrested in its Atlanta operation had been charged with or convicted of a crime. AJC
- Voluntary departure is cheaper and more humane. DHS says a paid self-deportation costs about $2,500 versus $18,000 for a removal, and shortens detention stays for those who choose it. Axios
- Labor-market effects for U.S. workers. The administration and DOL argue that tighter I-9 enforcement pushes employers toward lawful hiring and higher starting wages; this is the least documented of the claims, and no in-period independent study was found.
The case against
- Most detainees have no conviction. ICE's own detention statistics, as compiled by TRAC, show 70.6% of people in detention have no criminal conviction; about 40–48% have no pending charge either. Critics say the "worst of the worst" framing doesn't match who is actually held. TRAC · Kocher
- Due process is being thinned. Expedited removal skips the judge; Yajure Hurtado denies bond to a large class; there is no public defender in immigration court, and free counsel near Stewart has shrunk. WABE
- People are dying in custody. 37 deaths in fiscal 2026 per ICE's own death notices, the most on record; HRW ties the rise to overcrowding and medical understaffing. Two of the 2026 deaths were in Georgia facilities. HRW
- Citizens and legal residents get swept up. Wrongful detentions of U.S. citizens, including the Evanston case, and arrests at routine check-ins and courthouses. WTTW
- Cost and accountability. GAO (Congress's auditor) documented wasted detention spending; Congress funded ~41,500 beds and DHS is holding 65,000+ (ICE data) through emergency transfers; states are suing for the right to inspect private facilities. AP/PBS · Stateline
- Economic disruption. Farm, dairy and construction employers report they cannot fill vacated jobs; Georgia's Hyundai-site raid drew a $2 million claim. NPR · AJC
Where the numbers come from, and why both sides can cite the same data
Nearly every figure on this page, including the ones cited by critics, originates in the government's own records. Knowing the chain helps you judge any claim:
- ICE Detention Statistics — ICE's own biweekly spreadsheets (population, book-ins, conviction status, facility list). TRAC, Vera and Austin Kocher all compile this file; they add no data of their own. ice.gov
- ICE enforcement records obtained under FOIA — individual-level arrest and removal records released to the Deportation Data Project (UC Berkeley), which AP, NPR and Stateline report from. ICE's own methodology note explains that expedited removals processed by CBP are not counted as ICE removals, which is one reason totals differ between agencies. Deportation Data Project
- DHS Office of Homeland Security Statistics (OHSS) — the department's official statistics office: the monthly enforcement tables, the Enforcement Actions annual flow report and the Yearbook. These are the baseline definitions ("removal," "return," "book-in") every other source should be read against. Reporting in 2026 found the removals series had not been updated for the 2025 period and the ICE removals dashboard had not been updated since January 2025; where OHSS publishes, it is cited here first. NOTUS
- DHS press releases — headline totals (removals, "self-deportations") without the underlying tables. Shown on this page in the amber "reported but not checkable" tiles.
- Inspections and litigation — DHS Office of Inspector General and GAO reports, court orders, and NGO investigations (HRW, Vera, AIC). These describe conditions, not counts, and are attributed to their authors wherever they appear.
Put simply: supporters and critics are usually reading the same ICE spreadsheet. The disagreements are about definitions (conviction vs. pending charge; removal vs. return) and about which denominator to use. This page shows both wherever it can.
Where it's headed: the unresolved variables are the Supreme Court's rulings on third-country removals and Venezuelan TPS, the circuit split on bond hearings, whether Congress funds the detention level DHS is running, and whether removals can catch up with arrests. The data pages linked above update every two weeks.
How this page is checked
- A figure appears only if a person opened the linked page and saw it there. Rounded or secondhand numbers are marked approximate.
- TRAC compiles ICE's own detention spreadsheets every two weeks; Vera and the Deportation Data Project (UC Berkeley) use records obtained by FOIA. The official baseline definitions are OHSS's (DHS Office of Homeland Security Statistics); where OHSS has published for the period, it is cited first. DHS press-release totals are shown separately because the underlying tables have not been published.
- Advocacy and watchdog sources (HRW, Vera, AIC, ACLU) are used for conditions and litigation, never as the sole source for a count, and are named each time they appear.
- "No criminal conviction" (TRAC) includes people with pending charges; "no conviction and no pending charge" (Kocher) does not. Both are shown.
- Chart data lives in a public
data/series.jsonfile served with this site, with a source URL on every point.