US Immigration and Customs Enforcement (ICE) is a sub-agency of DHS. It deploys AI for case management, intelligence analysis, lead generation, biometric identification, and immigration-enforcement targeting, sourcing tools from commercial vendors including Palantir, Clearview AI, and OpenAI. It is the largest single contributor to growth in the DHS AI use-case inventory, and the federal agency whose AI deployments have generated the most litigation and civil-liberties criticism.
Scale of AI deployment
ICE reported 23 active AI software programs for immigration enforcement in May 2025 (Source: americanimmigrationcouncil.org). By the DHS inventory published in early 2026, ICE had added 24 new AI applications, covering tip processing, social-media and mobile-device data review, and facial recognition for identity confirmation; at least 23 applications across the DHS inventory use some form of facial recognition, face matching, or related biometric identification (Source: techpolicy.press).
Several programs listed as separate entries in the DHS inventory — ID scanning, mobile device analytics, video analysis, audio transcription — were marked inactive while their underlying functions persisted inside newer vendor platforms and field applications rather than being discontinued (Source: americanimmigrationcouncil.org).
AI uses and deployments
Lead generation and targeting
ELITE (Enhanced Leads Identification and Targeting for Enforcement) is a Palantir application that uses generative AI to help ICE officers extract information from records and warrants; the DHS inventory describes it as a tool to "extract accurate addresses and build usable enforcement leads." Reporting by 404 Media, which obtained the user guide, describes ELITE as generating a map populated with potential deportation targets and individual dossiers carrying an address "confidence score," drawing data from the Department of Health and Human Services and other government sources so that agents can identify neighborhoods for raids by data density (Source: techpolicy.press). Per the DHS AI use-case inventory, ICE began using ELITE in June 2025 (Source: npr.org). DHS's own characterization is narrower, stating that Palantir "provides solutions for investigative case management and enforcement operations" (Source: npr.org).
Biometric identification in the field
Mobile Fortify is a facial-recognition and fingerprint-matching application that compares biometric information captured in the field against agency records. It has been used by both CBP and ICE since May 2025, and appears twice in the DHS inventory, indicating that CBP operates the system and ICE accesses it (Source: techpolicy.press). A DHS document describes it as using "CBP's facial comparison or DHS's fingerprint matching to quickly verify subjects of interest during operations," and DHS has stated that the app "does not access open-source material, scrape social media, or rely on publicly available data" (Source: npr.org). Accounts of the underlying image pool put it at more than 200 million images held in DHS, FBI, and State Department databases (Source: law.vanderbilt.edu).
Reported use has expanded well beyond ports of entry. A lawsuit filed by the State of Illinois and the City of Chicago in January 2026 states that ICE had used Mobile Fortify in the field more than 100,000 times; the agency had previously limited facial recognition to ports of entry and investigations of child sexual exploitation. An internal DHS document obtained by 404 Media states that people cannot decline to be scanned and that facial scans are stored for 15 years regardless of citizenship or immigration status, while DHS has asserted to the Washington Post that "formal privacy oversight" applies with "strict limits" on data access, use, and retention (Source: law.vanderbilt.edu). 404 Media documented a case in which the app misidentified a woman twice during an immigration raid, notwithstanding ICE's position that its results are a "definitive" determination of immigration status (Source: techpolicy.press). Immigration lawyers told NPR of clients whose faces were scanned by Mobile Fortify and who were taken into custody even though the app failed to identify them (Source: npr.org).
Tip processing
A set of AI-enhanced tip-processing tools uses Palantir technology built on commercially available large language models to review, categorize, and translate incoming tips, including tips from the public (Source: techpolicy.press).
Social-media and communications monitoring
Procurement documents obtained by Wired show ICE planning to hire nearly 30 contractors to monitor Facebook, TikTok, Instagram, and YouTube around the clock, with the solicitation asking contractors to state how they would integrate AI into the work (Source: americanimmigrationcouncil.org). Voice, video, and text analysis feed into social-media and communications-monitoring platforms from Zignal Labs and PenLink; PenLink's Tangles product uses AI to assemble dossiers from a person's online presence, from social media to financial records (Sources: americanimmigrationcouncil.org; theguardian.com).
Procurement
Palantir is ICE's primary AI platform partner. Palantir's AIP, Gotham, and Foundry products are embedded in ICE operations for case management, targeting, and data fusion, and have been used to analyze driver's-license scans, extract information from seized phones, cross-search location data, and link records from federal, state, and commercial databases (Source: americanimmigrationcouncil.org). ICE also uses AI from OpenAI; The Information reported on January 29, 2026 under the headline "ICE Says It Uses AI From Palantir, OpenAI." OpenAI's specific role at ICE is less documented publicly than Palantir's.
Named contract figures include $30 million for Palantir's ImmigrationOS, which the vendor describes as offering "granular tracking" of immigrants including real-time monitoring of self-deportations; $4.6 million for iris-scanning smartphones from BI2 Technologies; and a $3.75 million Clearview AI facial-recognition contract characterized as ICE's largest purchase of that technology to date (Source: americanimmigrationcouncil.org). A separately reported ICE Homeland Security Investigations contract with Clearview AI is valued at $9.2 million, for software drawing on a proprietary database the company describes as holding more than 50 billion facial images (Source: immpolicytracking.org). The two figures are reported by different sources and are not reconciled here. Politico has reported ICE seeking to spend more than $300 million under the Trump administration on social-media monitoring tools, facial-recognition software, license-plate readers, and services locating where people live and work (Source: immpolicytracking.org).
A June 2026 report analyzing ICE and CBP contracts with eleven surveillance-technology companies traced spending back to 2013, when it stood under $50 million, and found steady growth with a larger increase over the preceding two years, driven principally by new contracts for Palantir and for Anduril, which supplies AI-powered surveillance systems, border towers, drones, and sensors (Source: theguardian.com). The same reporting identifies contracted categories including data brokers, analytics software, social-media scrapers, facial-recognition technologies, phone-hacking devices and spyware, external skip-tracing contractors, and autonomous border towers and drones. Other tools in the reported inventory include VeriWatch, a smartwatch-style device tracking migrants awaiting immigration proceedings (Source: theguardian.com).
Oversight and controversy
ICE's use of AI tools for immigration enforcement is among the more contested federal AI deployments. Civil-liberties advocates argue that AI amplifies enforcement capacity without proportional due-process protections, and have urged DHS to prohibit ICE from using Mobile Fortify in field operations, citing the intrusiveness of facial recognition, its potential to deter lawful protest, and persistent accuracy problems for women and people of colour — a concern sharpened by ICE's treatment of a positive face match as conclusive evidence of immigration status (Source: law.vanderbilt.edu).
A recurring line of criticism concerns the extension of tools justified for noncitizen enforcement to US citizens. Reporting has documented real-time facial scans of citizens, expanded purchases of commercial broker data, deployment of cell-site simulators and digital-forensics tools, and use of drones to surveil protests and enforcement operations, with ICE asserting authority to use these tools to monitor and investigate anti-ICE protest networks including US citizens (Source: immpolicytracking.org). NPR reported that DHS is using its surveillance apparatus to monitor, apprehend, and intimidate both people it seeks to deport and US citizens critical of its policies (Source: npr.org).
Litigation has followed. The ACLU sued over what it describes as a pattern of suspicionless stops, warrantless arrests, and racial profiling of Minnesotans by ICE and CBP, including through the use of facial recognition (Hussen v. Noem) (Source: techpolicy.press). The January 2026 Illinois and Chicago suit over Mobile Fortify is described above.
A related line of criticism concerns predictive algorithms in law-enforcement contexts more broadly. MIT Technology Review and Live Science (May 2025) reported on so-called "murder prediction" algorithms, with commentators arguing such predictive policing tools "echo some of Stalin's most horrific policies."
Relationships
- parent: DHS — Department of Homeland Security (AI Deployer)
- related: CBP — Customs and Border Protection (AI Deployer) (operates Mobile Fortify; ICE accesses it)
- deployed-by: Palantir Technologies, OpenAI, Anduril Industries
- related: AI and Surveillance, Surveillance Technology, AI and Authoritarianism, AI and Civil Liberties, Federal AI Adoption — Patterns and Tensions, Government AI Procurement
Revision note (2026-08-07 gap scan): Expanded from a 223-word stub carrying
confidence: highon a single source. All prior sourced content is retained. Built out from supporting web sources (DHS AI use-case inventory coverage, American Immigration Council, Tech Policy Press, NPR, Guardian, Vanderbilt Law, immpolicytracking.org);confidence: mediumper the gap-scan rule that an expanded page does not jump tohighwithout independent curation.