The scorecard
- What it is
- The 50 states plus D.C. and Puerto Rico — 52 jurisdictions, each scored 0 (no protection), 1 (partial), or 2 (full) on nine legal standards. 18 points possible.
- What it shows
- Where every state stands, standard by standard.
- Updated
- Methodology v3.4 · scored May 20 2026.
- Limits
- Enacted law only — no pending bills, no executive practice.
- Source
- Hand-audited statute record; every score traceable to a citation.
State bills
- What it is
- 328 state bills, direction-tagged.
- What it shows
- The momentum — who is adding limits, who is mandating cooperation.
- Updated
- Catalog through Jul 1, 2026.
- Limits
- A catalog, not a live tracker — status as of the last sweep; always confirm against the official text.
- Source
- State legislature sites, linked per bill.
287(g) & spending
- What it is
- Every local police or sheriff's department that has signed an agreement — called a “287(g)” — to help federal agents make immigration arrests, plus what the federal government has spent on detention in each state.
- What it shows
- Who signed up, and what the buildout costs.
- Updated
- ICE roster of Jul 14 2026 · spending through FY2025.
- Limits
- The roster is ICE's own publication; spending data trails the record.
- Source
- ICE.gov 287(g) roster; USASpending.gov.
The 287(g) roster
Every local agreement, by jurisdiction.
The county-attributed agreements behind the 2,127 nationwide total — the Task Force Model (street-level arrests) is the fastest-growing. State-agency MOAs and municipal signings ICE doesn’t tie to a county count in the total but have no row here. Filter by state and model; sort by agreements, population, or date.
Dates are the earliest signed agreement per jurisdiction, as ICE publishes them — a small number of roster dates appear mis-keyed and are shown as published or left blank.
Enforcement spending
Where the dollars land.
Reported ICE obligations by state for FY2025 — sortable by total and by dollars per resident.
A state with no row here means no obligation record was found in the federal data — not that it spent nothing. Absence from the table is a gap in the source, not a zero.
A full, contract-by-contract money trail is not shown here. The underlying award records list vendors, agencies, and dates — but not reliable dollar amounts — so a sortable money table awaits a clean obligation pull from USASpending. We won't publish dollar figures we can't stand behind.
Authority record
- What it is
- 95 documented federal actions and the checks that met them.
- What it shows
- What held, what didn't — the honesty ledger behind The Record.
- Updated
- Limits
- A curated selection, not a docket feed.
- Source
- Court records, agency documents, GAO — cited on every card.
The lead corpus
- What it is
- The raw lead corpus — the higher-signal subset of everything collected: entries that scored above our relevance threshold (the table header shows the live count).
- What it shows
- The volume and shape of reported enforcement, before curation — includes false positives; use the filters to narrow to real signal.
- Updated
- Continuous sweeps; the feed refreshes with each pipeline run.
- Limits
- Leads, not findings — public reporting, weight-scored, not individually verified. Nothing here is publishable without primary-source verification.
- Source
- Mixed press and primary feeds, credibility-weighted.
These are unverified leads. Automated sweeps of the public record, not yet fact-checked — a starting point for research, not a finding. The scored grades and the Federal Authority Record are the verified surfaces.
The “Strongest signal” view sorts by weight — our confidence that an item is real signal rather than noise, from 0.7 (lowest included) to 1.0 (highest). “Direction” tags whether a report shows federal force, state or local resistance, cost, governance, or public pressure.
Connected threads
- What it is
- 204 connected threads built from the corpus.
- What it shows
- Patterns over time — how separate reports connect.
- Updated
- Limits
- Built from the same lead corpus as Events; the same caveat applies.
- Source
- Each thread lists its own linked sources.
Threads are drawn from unverified leads. The patterns are machine-assembled from the lead corpus; each event links to its source so you can read the record yourself. The scored grades and the Federal Authority Record are the verified surfaces.