The name of the tool is ELITE: Enhanced Leads Identification and Targeting for Enforcement. It was built by Palantir Technologies for Immigration and Customs Enforcement. It looks, according to one description in court proceedings, “kind of like Google Maps” — a map interface that populates with potential deportation targets, each one marked with a confidence score indicating how likely the system is that it has the person’s current address correct.

That confidence score draws from a lot of places. Some of them are what you would expect from an immigration enforcement system — USCIS records, FBI data, DEA files, ATF records, travel records, visa documentation. Some are more surprising: phone records, school enrolment data, family relationship records, biometric traits captured in the field. And one source, confirmed in sworn court testimony and documented in EFF litigation, is the Department of Health and Human Services — which means, specifically, Medicaid.

EFF’s January 2026 report confirmed what its legal filings had been alleging: ELITE receives address data and identifying information drawn from Medicaid records covering approximately 80 million patients. Medicaid data is collected for one purpose — administering healthcare benefits for low-income individuals and families. It is now being used to find those individuals for deportation.

The Architecture of ELITE

Palantir has operated ICE’s Investigative Case Management (ICM) system since 2014 under a sole-source contract — no competitive bidding. Total contract value has exceeded $145 million following the integration of ImmigrationOS funding.

ICM is the underlying database layer: it consolidates records from across the federal government, including schooling records, family relations, phone records, biometric traits, criminal records, and tip submissions. ELITE is the operational interface that presents this consolidated data to field agents as an actionable targeting tool.

In April 2025, Palantir received an additional $30 million contract for ImmigrationOS — an AI-augmented layer on top of ELITE that provides “near real-time visibility” into self-deportations and visa overstays, with an initial prototype delivered in September 2025. The ImmigrationOS contract runs through September 2027.

ELITE also draws from Thomson Reuters’ CLEAR database — a commercial data broker product that aggregates public records, utility data, and financial information. The system additionally incorporates feeds from FALCON, Palantir’s tip-processing tool that has been used since 2012 to receive and categorise tips submitted to ICE’s tip line. Generative AI within the current ELITE system automatically summarises, categorises, and translates incoming tips.

The picture that emerges from the contract documents, court testimony, and leaked materials is of a system that consolidates health records, education records, family data, commercial data broker profiles, phone records, financial information, law enforcement records, and geolocation data into a single interface designed to help ICE agents find and arrest people.

The Leaked User Guide

A leaked ELITE user guide, reported by the EFF and reviewed by civil liberties organisations, contains an instruction that draws a direct line between the tool’s design and the pattern of arrests: operators are told they can disable filters to display all targets within a Special Operations dataset.

The filter-disable instruction undermines Palantir’s public position on ELITE. Palantir has consistently stated that ELITE is used for “prioritised enforcement” targeting individuals with removal orders or serious criminal histories. The leaked guide describes a mode of operation where those priority filters are turned off — where the system surfaces everyone in a dataset rather than those meeting stated criteria.

This is the technical architecture of a dragnet tool, not a targeted enforcement tool. The existence of priority filters does not prevent the tool from being used without them. The manual explicitly provides instructions for doing so.

What the Arrest Data Shows

Court documents and independent analysis of ICE detention records provide some empirical grounding for what ELITE-assisted operations look like in practice.

Nearly one in five ICE arrests during periods when ELITE-assisted sweeps were documented involved Latinx individuals with neither a criminal history nor a standing removal order. The individuals were arrested not because they met the stated criteria for enforcement priority but because they were present in an area that ELITE had identified as “target rich” for quota fulfillment.

The Medicaid data link makes this pattern especially damaging. People enrolled in Medicaid are, by definition, low-income individuals and families who sought healthcare. They provided their addresses to a government agency for the purpose of receiving medical coverage. The expectation of anyone filling out a Medicaid application is that the information will be used to administer benefits. The legal basis for repurposing it for immigration enforcement has been contested in federal court by the EFF, which sought injunctions in 2025 to block the data-sharing agreement between HHS and ICE.

Palantir claims ELITE maintains “indelible” audit logs tracking all operator activity. The EFF disputes the adequacy of that characterisation given the documented pattern of arrests, and argues that the existence of audit logs is irrelevant to the dragnet problem if the audit logs simply record that the filter was disabled.

What Palantir Says

Palantir published a human rights policy in 2021 that commits the company to assessing its products against UN Guiding Principles on Business and Human Rights. The EFF’s April 2026 analysis of that policy is titled “Palantir Has a Human Rights Policy. Its ICE Work Tells a Different Story.”

The ACLU has documented Palantir’s role in supporting deportation operations in materials published in 2026 covering the full range of surveillance tools ICE employs. Palantir’s position — that its tools are used for legitimate prioritised enforcement and that the audit infrastructure prevents misuse — has not been vindicated by the documented pattern of outcomes.

Palantir is a publicly traded company whose government revenue, particularly from US federal agencies, has grown substantially during the current administration. The ELITE contract predates the current administration and survived multiple transitions. The company has chosen not to exit the contract.

The Data Repurposing Problem

The Medicaid data channel represents the clearest example of what privacy scholars call function creep — the use of data collected for one purpose in service of an entirely different one. It is also a demonstration of what happens when data silos that were once administratively separate are connected through a common platform.

HIPAA, the federal health information privacy law, permits data sharing between federal agencies under specific conditions. The HHS-ICE data-sharing agreement has been argued by ICE’s legal team to fall within those conditions. The EFF argues the opposite. The courts have not definitively resolved the question.

The broader principle is not resolved by the legal outcome in any specific case. When health data, education records, family information, and commercial profiles can all be routed into a single enforcement interface, the practical privacy protections that sectoral laws were designed to provide — HIPAA for health, FERPA for education, various state laws for commercial data — lose their meaning. The silos exist in law; the integration exists in the database.

Primary sources: EFF Deeplinks (January 2026, April 2026), ACLU analysis, court filings in EFF v. HHS, State of Surveillance reporting (2026), Palantir contract documents via public records.