Federal Limits

The Nine Limits

Draw the line.

Nine limits a state can write into its own law. Each one closes a channel through which federal power expands unchecked. It was proven out against immigrants first — now it’s the template for reaching everyone else. Each limit is drawn from a law a state already enacted, built on what survived in court.

Scroll to draw it ↓

01
Resources

No state resources used

287(g) deputizations: ~135 → 2,127 in eighteen months.

Washington can’t force a state to carry out federal civil enforcement — so it started paying for it. A DHS salary-and-bounty program effective Oct 1, 2025 now funds local officers per arrest, and deputization agreements jumped roughly fifteenfold to 2,127 in eighteen months. Because paying isn’t compelling, the courts don’t reach it — only a state law declining to take part does.

What it does: bars the state’s own officers, money, equipment, and facilities from being used for federal civil enforcement.

Already law in 16 states + D.C.11 at full strength

What we propose

Bar any state or local agency from using money, staff, facilities, or equipment for federal civil enforcement — including honoring civil detainers without a judge’s warrant — while still complying with valid criminal warrants. Model: California SB 54 (§7284.6); Illinois Way Forward Act.

Who’s already done it

16 states + D.C. have a version; 11 at full strength — CA, CO, CT, DC, IL, MA, ME, NJ, NY, OR, WA. California’s SB 54 (2017) is the tested model — challenged in federal court and upheld.

On the record

  • DHS officer-salary & per-arrest payment program, effective Oct 1 2025
  • ICE 287(g) roster — 2,127 MOAs (Jul 14 2026)
  • Anti-commandeering: New York v. United States (1992); Printz (1997); Murphy v. NCAA (2018)

Status: mixed — courts struck the money-coercion vector; the paid-personnel vector is surging unchecked.

Full dossier: the state resources limit →

02
Contracts

No deputized local police

2,127 deputization deals — about 33% of Americans now live in a 287(g) county.

287(g) lets DHS deputize local police as federal civil enforcement agents. An executive order ordered maximum expansion, and signed agreements grew roughly fifteenfold to 2,127 — including the street-level model a federal court tied to systemic race-based policing in Melendres v. Arpaio. With federal audits paused, a state contract ban is the only durable off-switch; where no statute exists, courts have let the deals stand.

What it does: bars state and local agencies from entering, renewing, or maintaining any 287(g) agreement.

Already law in 13 states + D.C.9 at full strength

What we propose

Prohibit any state or local agency from entering, extending, or maintaining a 287(g) agreement of any kind, terminate existing ones within 180 days, and void those that don’t comply. Model: California Gov. Code §7284.9; Maryland HB 444 (2026).

Who’s already done it

13 states + D.C.; 9 at full strength — CA, DE, IL, MD, NJ, NM, NY, OR, WA. California’s §7284.9 is the upheld model; Maryland’s HB 444 (2026) is the cleanest 180-day wind-down.

On the record

  • ICE 287(g) roster — 2,127 MOAs (JEM 177 / TFM 1,422 / WSO 528), Jul 14 2026
  • Executive Order 14159 §11 (Jan 20 2025)
  • Melendres v. Arpaio (D. Ariz., 2013) — race-based policing under a task-force agency

Status: surging and largely unchecked federally; a statutory ban holds where enacted, fails where absent.

Full dossier: the deputization contracts limit →

03
Sensitive locations

Off-limits: schools, hospitals, courthouses

+22% student absences after raids — the protection erased by a single memo.

From 2011, the only thing keeping enforcement out of schools, hospitals, and courthouses was a federal self-restraint — rescinded in one memo on Jan 20, 2025. A peer-reviewed study found a 22% jump in student absences across districts serving 500,000+ children after raids; hospital and school incursions are now documented across at least six metro areas. Court protections are partial and reversible — only a state property law is durable.

What it does: bars state and local facilitation of enforcement at protected places and requires a judge’s warrant before entry to their non-public areas.

Already law in 14 states + D.C.8 at full strength

What we propose

Protect six location types (courthouses, K-12 schools, hospitals and clinics, childcare, places of worship, workplaces); require a judge’s warrant, true exigency, or written consent before entry to non-public areas; add notice, documentation, and a private right to sue. Model: California SB 54 (§7284.4) + AB 49/AB 495; New York’s Protect Our Courts Act.

Who’s already done it

14 states + D.C.; 8 at full strength — CA, CO, CT, IL, MD, NJ, NY, OR. California builds the fullest list; New York’s Protect Our Courts Act anchors the courthouse piece (upheld; U.S. v. New York dismissed Nov 17 2025).

On the record

  • DHS rescission memo, Jan 20 2025 (protected-areas policy ended)
  • Dee, PNAS, Nov 2025 — +22% student absences after raids
  • U.S. v. New York (N.D.N.Y.) — dismissed Nov 17 2025

Status: mixed — the federal rescission is in force and court shields are partial; the state property lever is the only restraint not dependent on federal forbearance.

Full dossier: the sensitive locations limit →

04
Private detention

Limit for-profit detention

Detention +71%, facilities +78% — 39 deaths in custody, the most since 2003.

For-profit detention is the engine of the surge: the bed count is the removal capacity. After the federal detention-oversight office was dissolved in March 2025, the number of people held rose 71% and the number of facilities holding them rose 78% — backing a 100,000-bed target, with $2.1 billion to one contractor and $653 million to another in a single year. The more beds they fill, the more they are paid.

What it does: bars private contractors from operating civil detention in the state, and withholds the state permits, funding, and leases that enable it.

Already law in 6 statesonly 2 at full strength

What we propose

Withhold the state’s own permits, funding, leases, and zoning from for-profit civil detention — the lever courts have left standing — and bar private operation where it holds, with a phase-out and a severability clause so the property prongs survive if the operating ban is enjoined. Model: Illinois 730 ILCS 141 (federal challenge dismissed); Maryland HB 1018 (zoning).

Who’s already done it

6 states have acted; only Illinois and Oregon at full strength. Illinois’s 2019 Moratorium Act is the durable model — its federal challenge was dismissed; California’s flat ban was struck, which is exactly why the Illinois design matters.

On the record

  • GAO-26-108886 (Jun 9 2026) — detention +71%, facilities +78%
  • GAO-25-107580 — detention-oversight office dissolved March 2025
  • Illinois 730 ILCS 141 (2019) — federal challenge dismissed Jan 2024

Status: mixed — flat bans have been struck; the durable slice is the state’s own permits, funding, and leases (the Illinois model).

Full dossier: the private detention limit →

05
Information firewall

Protect personal records

67M+ voter records bulk-scanned; 292,114 searches of state driver records in a year.

Starting in 2025, the federal government moved to fuse state-held records — voter rolls, DMV files, tax data — into a single surveillance system, with deportation as the current use case and no legal wall stopping the next one: 31 lawsuits across 30 states over voter rolls, a rebuilt SAVE system scanning 67 million-plus registrations, an IRS deal exposing roughly 47,000 taxpayers, and 292,114 searches of DMV records. Courts struck the worst of it — but those are reversible orders; only a state firewall law is durable.

What it does: bars state agencies and local data-holders from handing residents’ voter, DMV, school, health, and licensing data to federal civil enforcement without a judge’s warrant.

Already law in 16 states + D.C.7 at full strength

What we propose

Firewall three or more of five data domains (voter, DMV, school, health, licensing) with hard-denial language, a judicial-warrant exception, public logging of refused requests, and a SAVE opt-out. Model: Colorado SB 24-182; California Veh. Code 1801.1; Illinois Way Forward Act.

Who’s already done it

16 states + D.C. have firewalls; 7 at full strength — CA, CO, CT, IL, MD, NJ, OR. California and Illinois firewall all five domains.

On the record

  • League of Women Voters v. DHS — SAVE overhaul struck, Jun 22 2026
  • DOJ voter-roll suits — 31 across 30 states; 6th Cir. affirmed against, Jun 24 2026
  • ICE searches of state driver records — 292,114 via Nlets in a year (U.S. Senate letter to governors, Nov 2025)
  • IRS-ICE data deal — struck twice (D.D.C.; D. Mass.)

Status: holding in court — but via reversible litigation, not durable statute. Oklahoma already settled; a state law is the answer.

Full dossier: the data firewall limit →

06
Warrant

Require a judge’s warrant

0 of 36 arrest reports complied with a federal court’s injunction — arrests run on agency forms, not warrants.

Federal civil arrests run on forms an agency supervisor signs, not a neutral judge. Four federal courts enjoined the warrantless practice and three found defiance; in Colorado, not one of the 36 arrest reports produced under the injunction complied — “It is as if this court’s order never happened,” as the ACLU attorney on the case put it. A state judicial-warrant law is the self-executing answer.

What it does: requires a judge’s warrant before any state or local actor detains or transfers a person for federal civil enforcement; agency forms and detainers don’t qualify.

Already law in 14 states9 at full strength

What we propose

Bar state and local actors from detaining or transferring anyone into federal civil custody without a judge-signed warrant; state plainly that an I-247 detainer or I-200 administrative warrant does not qualify; shield agencies that decline. Model: California SB 54 (§7284.6); Massachusetts Lunn v. Commonwealth.

Who’s already done it

14 states; 9 at full strength — CA, CO, CT, IL, MA, ME, NJ, NY, OR. California’s SB 54 is the statutory model; Massachusetts reached the same bar through its high court (Lunn, 2017).

On the record

  • Ramirez Ovando v. Noem (D. Colo.) — noncompliance finding May 12 2026
  • Castanon Nava v. DHS — consent decree extended Oct 7 2025; upheld on appeal Dec 11 2025 and May 5 2026
  • 8 U.S.C. 1357(a)(2); Galarza v. Szalczyk (3d Cir. 2014)

Status: acute — the warrantless lane is the federal mainline; courts engaged, defiance documented. The state statute is the durable fix.

Full dossier: the judicial warrant limit →

07
Transparency

Public cooperation record

Federal detention data went dark up to 56 days — while detention rose 71%.

You can’t contest what you can’t see. Federal civil enforcement’s congressionally-mandated detention data went dark for up to 56 days in early 2026, and its public dashboard is roughly eighteen months stale — all while detention rose 71%. With federal reporting defied and 11 withheld records requests now in litigation, a state reporting law turns the lights back on — building a public record that survives when the federal one goes dark.

What it does: requires public reporting, by statute, on the state’s own interactions with federal civil enforcement.

Already law in 6 states + D.C.only 3 at full strength

What we propose

Each agency that interacts with federal civil enforcement reports quarterly to a state authority and posts publicly within 30 days, with an annual statewide aggregate — covering transfers, detainers, access requests, and contracts; non-reporting is subject to attorney-general action. Model: California Gov. Code §7284.10; Colorado SB 21-131.

Who’s already done it

6 states + D.C. have qualifying reporting; 3 at full strength — CA, IL, NJ. California’s Values Act requires an annual public report on cooperation from the attorney general.

On the record

  • Consolidated Appropriations Act 2020 (P.L. 116-93) — statutory publication mandate
  • GAO-24-106233 — DHS non-concurred; unimplemented as of May 2026
  • GAO-26-108886 — detention +71% over the same window

Status: mixed — the federal mandate is intact but actively defied; the state reporting lever (CA, IL, NJ) is the operative check.

Full dossier: the transparency limit →

08
Remedies

Hold enforcement accountable

After Goldey v. Fields (June 2025), no federal remedy is left when enforcement crosses the line.

A limit with no remedy is a suggestion. In 2025–26 the federal paths to accountability closed at once: the Supreme Court barred the core damages claim (Goldey v. Fields, Jun 30 2025), tort immunity walls off the rest, and the DHS oversight offices were gutted. Two U.S. citizens were killed in enforcement operations — and their families hit the same immunity wall. A state right to sue is the remedy that’s left.

What it does: makes the state’s limits enforceable — through a private right to sue and/or attorney-general authority reaching local law enforcement.

Already law in 12 states9 at full strength

What we propose

Create a private right of action (damages, fees, and a statutory minimum per unlawful detention) plus attorney-general authority to investigate and penalize, with a non-retaliation clause. Model: Connecticut HB 7212 (2025); the Illinois Bivens Act (PA 104-0440).

Who’s already done it

12 states have a mechanism; 9 at full strength — CA, CO, CT, IL, MD, NM, NY, OR, WA. Connecticut and Oregon do both (sue + AG); Illinois’s Bivens Act recreates the very federal remedy the Supreme Court withdrew.

On the record

  • Goldey v. Fields, 606 U.S. 942 (Jun 30 2025) — damages claim barred
  • Illinois Bivens Act (PA 104-0440), eff. Dec 9 2025
  • Connecticut HB 7212 (2025), eff. Oct 1 2025

Status: contested — federal remedies closed; states are recreating them. Whether a state may is being tested in U.S. v. Illinois.

Full dossier: the remedies limit →

09
Agent identification

Identified agents, recording allowed

~1,618 arrests in 16 days by agents “many lacking identifying information.”

Unidentifiable federal agents — masked, in plain clothes, in unmarked cars — are becoming the default mode of enforcement, and that anonymity is by design: it strips the public’s ability to know who is detaining them or hold anyone accountable, in any federal operation. A U.S. senator’s office documented roughly 1,618 arrests in one metro area over sixteen days by agents “many lacking identifying information,” and a federal court found the masking served “to terrorize Americans into quiescence.” With federal oversight dismantled, a state ID-and-recording law restores the minimum proof of who did what.

What it does: requires the state’s own officers to show ID and not mask on duty during cooperation, keeps cameras recording, and protects the public’s right to record.

Already law in 6 states5 at full strength

What we propose

Require visible ID (name or number, agency, badge), bar non-medical face coverings (narrow tactical, PPE, and religious exceptions), keep body cameras recording during federal cooperation, and protect the public right to record — with civil enforcement. Model: Oregon HB 4138; Connecticut SB 397.

Who’s already done it

6 states have ID/mask/recording provisions; 5 at full strength — CA, CT, NJ, NY, OR. Oregon’s HB 4138 (Apr 2026) is the cleanest template; New York’s MELT Act is the live federal test.

On the record

  • AAUP v. Rubio (D. Mass., Sep 30 2025) — masking finding against the government
  • LA Press Club v. Noem (9th Cir., Apr 1 2026) — injunction affirmed on First Amendment retaliation grounds; terms narrowed on remand
  • Oregon HB 4138 (Apr 9 2026); Connecticut SB 397 (May 4 2026)

Status: live — the state-officer slice survives; the federal-officer reach is being tested (New York’s MELT Act).

Full dossier: the agent identification limit →

Nine lines.
One model bill.

Your legislature can pass it this session. But the average state has locked in about two of the nine — while six states have drawn nearly all of them.

Read the model bill → Or see where my state stands →