The Questionable Legality of ICE's Palantir Elite System

Immigration and Customs Enforcement (ICE) uses a Palantir-developed tool called Enhanced Leads Identification & Targeting for Enforcement (ELITE) to populate maps with potential deportation targets, generate dossiers, and assign confidence scores to individuals' addresses, according to internal ICE materials reviewed by 404 Media in January. The reporting conflicts with the Department of Homeland Security's characterization of ELITE as producing "normalized address data" that is not a "principal basis" for legally significant decisions, while subsequent reporting and sworn officer testimony indicate ICE also uses ELITE to identify "target rich" areas with a dense concentration of people with an "immigration nexus." The article analyzes unresolved legal questions about whether ELITE enables suspicionless, profile-based, or sweep-style enforcement, whether targets can interrogate the basis for their arrest or detention, and whether ICE's address-normalization claim matches field use, noting the public record still lacks information on ELITE's data sources, accuracy, contractual status, and human-review safeguards.

From Records to Raids: The Questionable Legality of ICE’s Palantir ELITE System Sarah Engelberg-Nolan, Emma Haberman, Nicole Brosnan / Sep 21, 2026 In January, 404 Media reported https://www.404media.co/elite-the-palantir-app-ice-uses-to-find-neighborhoods-to-raid/ that Immigration and Customs Enforcement ICE uses a Palantir-developed tool called Enhanced Leads Identification & Targeting for Enforcement ELITE to support immigration enforcement operations. According to internal ICE materials reviewed by 404 Media, ELITE can populate maps with potential deportation targets, generate dossiers on individuals, and assign a confidence score to a person’s current address—all using data drawn from various government databases, private data brokers, and the open internet. In conflict with this reporting is DHS’s own characterization https://www.techpolicy.press/dhs-ai-surveillance-arsenal-grows-as-agency-defies-courts/ of the ELITE system as a tool that produces “normalized address data” and is not supposed to serve as a “principal basis” for legally significant decisions, taking it out of the category of a “high impact” AI system that would be subject to more exacting standards and oversight. Subsequent reporting and sworn officer testimony indicate that ICE not only uses ELITE to look for individual targets, but it also uses the tool to identify “target rich” areas believed to contain a dense concentration of people with an “immigration nexus.” ELITE transforms administrative, public, or private data—much of which is mundane—into operational intelligence for field immigration enforcement. But the public record is still bereft of information about ELITE’s precise data sources, accuracy, contractual status, and human-review safeguards, as well as how agents are trained to use it and its actual role in investigation, stops, arrests, detention decisions, and removal proceedings. This article analyzes what is known https://www.404media.co/here-is-the-user-guide-for-elite-the-tool-palantir-made-for-ice/ about the system to identify the most important legal questions raised by ELITE: whether it enables suspicionless, profile-based, or sweep-style enforcement; whether people targeted through the system can meaningfully interrogate the basis for their arrest or detention; and whether ICE’s formal claim that ELITE is merely address-normalization software matches how the tool is actually used in the field. 1: Development and deployment of ELITE What exactly did ICE procure: ELITE, ImmigrationOS, or a broader Palantir enforcement capability? The public record does not yet answer this cleanly. ELITE appears to be a Palantir-developed application, use case, or adjacent capability within ICE’s broader Palantir enforcement stack. That stack includes Palantir’s long-running Investigative Case Management ICM system and the ImmigrationOS overlay, which was publicly described as including “Targeting and Enforcement Prioritization, “Self-Deportation Tracking,” and immigration-lifecycle logistics. The overlap between those workstreams and ELITE’s reported functions is substantial, but the precise contractual relationship remains unresolved. The legality of the procurement depends on what ICE actually bought, how it described the work, and whether the agency bundled new enforcement-targeting capabilities into an incumbent Palantir contract. Did DHS follow relevant laws and procedures in procuring the ELITE software? Maybe, but the better question is narrower: whether ICE properly used a noncompetitive or urgent-and-compelling justification to expand Palantir’s role in immigration enforcement, and whether ELITE falls within that expansion. ICE awarded Palantir a sole-source ImmigrationOS contract, justified on urgency and Palantir’s unique capacity to deliver the system, invoking a Federal Acquisition Regulation FAR exception https://iptp-production.s3.amazonaws.com/media/documents/2025.04.17 ICE Limited Sources Justification for 70CTD022FR0000170.pdf that allows DHS to circumvent the competitive procurement process. If ELITE was built under, added to, or functionally enabled by that work, then the procurement question becomes whether ICE legally treated Palantir’s incumbent status as a reason to avoid competitive bidding. DHS may have violated the Competition in Contracting Act https://www.congress.gov/bill/98th-congress/house-bill/5184 CICA by not seeking competing company’s bids for the development of this software. That question cuts both ways. ICE would likely argue that urgent enforcement priorities, interoperability, and Palantir’s long experience with ICE systems made a noncompetitive award lawful. The harder counterargument is that vendor lock-in cannot automatically become its own justification. If an agency’s existing dependence on Palantir is what makes Palantir the only practical supplier, then the urgent-and-compelling rationale risks converting past procurement choices into indefinite sole-source control over new enforcement infrastructure. Did ICE comply with federal privacy, matching, and data-governance rules when developing ELITE? Potentially not. ELITE software is reported to aggregate address data from the Department of Health and Human Services HHS and cross-reference it with ICE data, including the Enforcement Integrated Database. That raises serious Privacy Act, Computer Matching and Privacy Protection Act https://www.congress.gov/crs-product/R47325 CMPPA , system-of-record notice, privacy impact assessment, and routine-use questions. The CMPPA requires that where an agency is using a person’s information in a “matching program,” such as here where ELITE provides an address confidence score by aggregating data, the individual must be notified and be permitted to verify their information. This step must occur before any “adverse action” is taken against the individual https://www.congress.gov/crs-product/R47325 . Further, CMPPA requires a formal agreement between the agencies https://www.congress.gov/crs-product/R47325 . But the existing public record is not strong enough to confirm whether ELITE is a statutorily covered “matching program.” Further, there is no clear indication that people targeted via the ELITE application were notified that their data was being matched, nor were they provided with an opportunity to verify such information. There is also no public record of a formal matching agreement between the agencies. At a minimum, ELITE exposes a gap between data-sharing law and operational enforcement reality. If ICE is using HHS-derived address records or other interagency data to generate leads, then the key legal questions are: which systems of record supplied the data; what routine uses or data-sharing agreements authorized the transfer; whether the system is a “matching program”; and whether the records used were accurate, relevant, timely, and complete. Does ICE’s use of ELITE violate the federal AI-governance requirements or DHS’s own classification of ELITE? Maybe. The Trump administration announced its intention to expand AI use in OMB Memorandum M-25-21 https://www.whitehouse.gov/wp-content/uploads/2025/02/M-25-21-Accelerating-Federal-Use-of-AI-through-Innovation-Governance-and-Public-Trust.pdf . But such expansion includes “risk management practices” when an AI is high impact—meaning it “could have significant impacts when deployed.” Here, DHS has determined https://fedscoop.com/dhs-ai-inventory-mobile-fortify-palantir/ that ELITE is not high impact, meaning its use does not require a risk management practice. It’s not clear whether this determination may be challenged, but without designating ELITE as a high impact AI, it is not likely that DHS is required by OMB Memorandum M-25-21 to engage in risk management practices. The legal question, then, is whether DHS’s low-impact or non-principal-basis characterization matches field reality. If ELITE merely cleans address data later verified by agents, DHS’s classification is defensible. But credible reporting suggests that description is misleading at best. If ELITE, as reporting suggests, materially steers stops, sweeps, home visits, arrests, or detention-related decisions, challengers can argue DHS understated the system’s impact and failed to comply with mandatory disclosure, safeguard, testing, validation, monitoring, and human-review requirements. The OMD AI-governance memo, however, does not on its own create a private right of action; even if it is being violated, that is not sufficient to bring a cause of action. 2. Feeding data into the tool What data feeds ELITE, and was it lawfully collected, shared, and repurposed for immigration enforcement? ELITE combines HHS-derived address data and combines it with ICE data. That ICE data may include information ICE already possesses and claims to have lawfully obtained, including records incorporated from other agencies, prior enforcement systems, or commercial vendors. But the full source architecture is unknown: the public record does not establish whether additional data streams—such as DMV data, ALPR data, mobile-location data, tipline data, or brokered records—feeds directly into ELITE. Given that ELITE appears to be built off of or at least interoperate with Palantir’s existing systems, including ImmigrationOS, which does draw from these data streams, there is reason to believe ELITE can and is accessing this data. Is the Fourth Amendment implicated when the information used in ELITE was gathered in the first instance? Maybe, depending on the source. Where ICE collects data directly, the Fourth Amendment demands judicial process—generally a warrant or legitimate application of an exception to the Fourth Amendment—if the data collection is deemed a search. Per Katz v. United States https://supreme.justia.com/cases/federal/us/389/347/ , a search occurs when the government violates a reasonable expectation of privacy. United States v. Jones https://www.law.cornell.edu/supremecourt/text/10-1259 held that continuous monitoring of location data, even in public spaces, does violate a reasonable expectation of privacy. The thorny question is whether individuals retain a reasonable expectation of privacy in data they voluntarily hand to third parties—traditionally, courts have held they do not, because of the “third-party doctrine.” However, recent cases have signaled a narrowing of the third-party doctrine, which opens the door to challenges to the ELITE system if it uses data obtained from other sources without first obtaining a warrant. Broadly speaking, the Fourth Amendment does not prohibit the government from purchasing brokered data without judicial process, though many argue this is a circumvention of the Constitution https://www.aclu.org/news/privacy-technology/dhs-is-circumventing-constitution-by-buying-data-it-would-normally-need-a-warrant-to-access . But some forms of location-data collection from third-parties may be searches requiring judicial process. Carpenter v. United States https://epic.org/documents/carpenter-v-united-states-2/ :~:text=In%20Carpenter%20v.,search%E2%80%9D%20under%20the%20Fourth%20Amendment. held that when the government obtains a large volume of location data over a long enough period of time from a third-party cell-service provider, they have conducted a search under the Fourth Amendment and therefore must abide by its requirements. Other court cases have also suggested that movement tracking through third-party ALPRs should https://cases.justia.com/federal/appellate-courts/ca7/20-2352/20-2352-2021-07-14.pdf?ts=1626289227 trigger Fourth Amendment protections. The recent SCOTUS decision in Chatrie v. United States https://www.supremecourt.gov/opinions/25pdf/25-112 0am4.pdf strengthens the argument that even a small handful of location data obtained using a geofence warrant can trigger individualized Fourth Amendment protections. So, if ELITE truly relies on ordinary address records lawfully collected by ICE or lawfully obtained from HHS, the Fourth Amendment argument is harder. If it does indeed rely on third-party data, such as ALPR or mobile-location data, the argument is stronger. Does indefinitely retaining location or identity data in government databases amount to a Fourth Amendment seizure? Potentially, but this is a developing argument. The theory is that even if each data point was lawfully collected, indefinite retention and aggregation may give the government a durable surveillance archive that is materially different from the search, thereby meaningfully interfering with a person’s possessory interest or rightful control over their personal information. Courts more commonly analyze government access to aggregated data as a search, not a seizure, so this is not the strongest or most settled Fourth Amendment theory. But it is a colorable argument where the government maintains the ingredients for personal dossiers indefinitely without individualized legal basis. For ELITE, the key factual question is whether ICE is merely storing discrete records, or maintaining a searchable, cross-source location profile capable of revealing where someone lives, works, receives services, or is likely to be found. Is querying ELITE to verify an address, generate leads, or identify targets itself a Fourth Amendment search? Even if the initial collection, integration, and indefinite retention of data is legally justified, it is an open question whether the algorithmic query or output creates a new constitutional event: the government uses a system to synthesize scattered records into an address-confidence score, target map, or enforcement lead. The “mosaic theory” laid out by Justice Sonia Sotomayor in Jones https://supreme.justia.com/cases/federal/us/565/400/ has not been formally adopted, but it has been persuasive, which supports the idea that the Fourth Amendment has something to say when integrating multiple data sources https://scholarship.law.umn.edu/mlr/91/ “reveal s information that, in the absence of aggregation, the government could only access by conducting a search or seizure, the extraction of that information should be subject to constitutionally based limits.” Constitutional concerns about data aggregation and retention liken it to warrantless searches https://www.aclu.org/news/privacy-technology/new-records-detail-dhs-purchase-and-use-of-vast-quantities-of-cell-phone-location-data allowing for fishing expeditions. The government would likely argue that querying a database is not a search if the underlying records are lawfully available. The counterargument is that ELITE’s output may reveal something no single record reveals on its own, making the generation of an output a separate action that potentially violates a reasonable expectation of privacy absent a warrant. 3. Use of the output: reasonable suspicion, probable cause, and due process Does an ELITE-assisted stop satisfy the Fourth Amendment’s reasonable suspicion requirement? Maybe. A tool does not itself “meet” reasonable suspicion; an officer must have specific, articulable, individualized facts that justify the stop. Per United States v. Brignoni-Ponce https://supreme.justia.com/cases/federal/us/422/873/ , officers may not rely on race or ethnicity alone to justify an immigration stop, but that those features can be part of a totality of the circumstances evaluation of reasonable suspicion. The legal questions become whether the system’s recommendation of a target can itself meet the reasonable suspicion standard, in that it its determinations are based individualized, specific, articulable facts that are not based on race of ethnicity; whether the system’s recommendation of a target can play a role in conjunction with other factors to establish reasonable suspicion, even if its recommendations over-index to race or ethnicity; and whether operations on the ground use ELITE as the functional sole basis for a stop even while claiming it is not the “principal basis” of that legal decision, which would be illegal if the system outputs fail to meet the reasonable suspicion standard on their own. Recent litigation complicates the analysis but does not eliminate the individualized-suspicion requirement. In Noem v. Vazquez Perdomo , the Supreme Court granted an emergency stay of a district-court injunction that had restricted profile-based immigration stops in Los Angeles. The Court did not issue a merits opinion. But, Justice Kavanaugh’s separate concurrence https://www.supremecourt.gov/opinions/24pdf/25a169 5h25.pdf suggested that reasonable suspicion can be established by a combination of factors including presence at locations such as bus stops, car washes, day-labor pick up sites; the type of work a person appeared to do; speaking Spanish or English with an accent; and apparent race or ethnicity. Commentators and litigants have treated this concurrence as permitting stops based primarily on apparent ethnicity or race—giving rise to the phrase “Kavanaugh stops”—although Kavanaugh appears to have distanced himself from that reading. For ELITE, the legal questions hinge on a fundamental inquiry: how much does the system operationalize broader proxies for race and ethnicity—neighborhood, worksite, language, prior immigration contact, or “immigration nexus”—in generating target leads or identifying hot spots. If its algorithmic outputs are based predominantly on impermissible considerations of race or ethnicity, ICE agents’ reliance on the system to justify stops is on shakier legal grounds. Does ICE’s use of ELITE to engage in neighborhood “sweeps” violate the Fourth Amendment? Maybe. Sworn testimony described ELITE as helping officers identify “target-rich areas where arrests were more likely. But area-level suspicion is not individualized suspicion and cannot, on its own, justify stops. If ICE uses ELITE to decide whether to deploy agents, then agents stop people merely because they are present in the area or fit a broader profile, the decisions may not be sufficiently supported by specific and articulable facts of individualized reasonable suspicion. Does ELITE provide a lawful basis for warrantless civil immigration arrests? Not by itself. A reasonable suspicion stop and a civil immigration arrest are different events. Even if ELITE helps ICE locate someone or conduct an immigration stop legally, ICE still needs a lawful basis to escalate that encounter into an arrest. Under the Immigration and Nationality Act https://www.uscis.gov/laws-and-policy/legislation/immigration-and-nationality-act INA , ICE may arrest someone if the officer has 1 “reason to believe” the person is in the US in violation of immigration law; and 2 is “likely to escape before a warrant can be obtained.” ELITE may help satisfy the first prong, but an address-confidence score does not establish likelihood of escape and may even cut the other way if ICE has a tool that can tell them with relative confidence where the person can be found. Does ICE’s use of ELITE violate due process under the Fifth Amendment? Potentially, but the claim should be tied to a specific deprivation. Because ICE’s use of ELITE is occurring in a context where the public is not aware of the exact data inputs, it is increasingly difficult for people who are targeted by the ELITE application to contest their detention. But, opacity alone is not automatically a due process violation. Due process, per the Mathews v. Eldridge https://supreme.justia.com/cases/federal/us/424/319/ balancing test, requires notice and a meaningful opportunity to be heard before the government deprives an individual of a right or privilege. Due process is most clearly implicated when ELITE-derived information contributes to detention, bond, charging, removal proceedings, or other liberty-affecting decisions and the person is left without recourse to challenge the deprivation. Further, due process under Mathews v. Eldridge requires that the risk “of an erroneous deprivation” of a “private interest” be weighed. ICE agents themselves have indicated ELITE is not wholly reliable https://goldman.house.gov/media/press-releases/goldman-wyden-velazquez-demand-answers-ice-use-palantir-developed-technologies ; it can be wrong even when it reports high confidence and officers are supposed to perform additional checks, which there is reason to believe they may not do https://www.404media.co/ices-facial-recognition-app-misidentified-a-woman-twice/ , given they treat outputs from similar identification platforms like Mobile Fortify that are also error-prone as “definitive.” If those checks are not done, documented, or disclosed, ELITE may make it practically impossible for a person to challenge stale, mistaken, or unverified data to vindicate their Fifth Amendment rights. 4. Procedural and doctrinal barriers to raising issues in court Who has standing to challenge ICE’s use of ELITE in immigration arrests? Standing is a constitutional requirement to proceed in court https://supreme.justia.com/cases/federal/us/504/555/ , and requires that a person suffer 1 an injury in fact, that is 2 caused by the defendant, and 3 is redressable by judicial intervention. A person stopped, arrested, detained, or subjected to a home encounter because of ELITE may have a concrete injury. But, proving causation may be difficult because ICE is not likely to disclose whether ELITE was used, what it showed, or whether agents independently verified output. Systemic suits—often the route impact litigation pursues—face additional barriers: plaintiffs seeking forward-looking relief must show a real likelihood of future injury, which will be challenging without access to ELITE system specifications, use procedures, audits, logs, and error reports. Additionally, immigration-related claims may be channeled through removal proceedings, habeas, petitions for review, or limited district-court actions—some of these fora do not confer on immigrants the same constitutional and civil procedure protections available in traditional court proceedings. In practice, the first litigation battles may have to focus on plaintiffs obtaining enough records to prove ELITE’s role in enforcement. Can ELITE-related violations lead to suppression, termination, or evidentiary relief in removal proceedings? Sometimes, but this can be a steep hill to climb. In civil removal proceedings, unlawful stops and arrests do not automatically lead to suppression of evidence. In INS v. Lopez-Mendoza , the Supreme Court held that an unlawful immigration arrest does not automatically make the government’s evidence unusable or require the case to be dismissed. Suppression is generally available only for especially serious constitutional violations, such as of the Fourth Amendment, or systemic unlawful practices, or the serious compromise of evidence reliability. Termination, or the dismissal of a removal case, is even harder to obtain, usually requiring that either the government cannot prove removability without the tainted evidence or that the violation caused serious prejudice. More granular information about ELITE may help show either unreliability of evidence used in proceedings or a broader unlawful pattern of enforcement. May organizations raise constitutional challenges under Section 1983 and Bivens ? Bivens ? Not likely. Section 1983 is a civil rights statute created in the wake of the Civil War that created a cause of action against state and local actors who violate the federal constitutional rights of others—not federal actors, such as ICE agents. In Bivens v. Six Unknown Named Agents https://supreme.justia.com/cases/federal/us/403/388/ , the Court expanded Section 1983 to create a cause of action against certain federal actors where such violations of federal constitutional rights were particularly egregious. Precedent indicates that this judge-made remedy is quite limited, making it unlikely that individuals can pursue recourse against ICE using Bivens actions. Bivens is also a cause of action against a federal officer as an individual; however, it may be difficult to identify the right defendant when it is unclear whether the ELITE-driven constitutional injury arose during data collection, system development, querying, or the use of outputs. Routes to challenge ELITE that are on firmer ground are more procedural: immigration-court suppression of evidence or subpoenas for more information about the system and its use; habeas or bond challenges to detention invoking the use of ELITE; FOIA and Privacy Act requests for government and contractor records related to ELITE; Administrative Procedures Act or equitable claims to challenge executive action in its procurement and use of ELITE; Federal Torts Claims Act suits against specific federal officers for wrongful acts or negligence on duty; and preservation demands for ELITE outputs, audit logs, source-data lineage, and officer communications. Conclusion There are important legal considerations implicated by ICE’s use of ELITE. The impact of this software—which gives armed agents the coordinates to target immigrants for detention in raids that have turned violent, sparking demonstrations and civil unrest—warrants further exploration, research, and examination. Here we aim to motivate continued discussions among the legal and policy community about the constitutional and regulatory questions raised. Authors Sarah Engelberg-Nolan https://www.techpolicy.press/author/sarah-engelberg-nolan Emma Haberman https://www.techpolicy.press/author/emma-haberman Nicole Brosnan https://www.techpolicy.press/author/nicole-brosnan