The Department of Homeland Security (DHS) is a federal deployer of AI concentrated in law-enforcement and border-security use cases, spanning its component agencies including Immigration and Customs Enforcement (ICE), Customs and Border Protection (CBP), and the Cybersecurity and Infrastructure Security Agency (CISA). DHS maintains a publicly available AI Use Case Inventory Library cataloging AI systems in use across the department. Law enforcement is the largest single category of DHS AI use, and biometric identification — facial recognition in particular — is the most common technique within it.
The AI Use Case Inventory
DHS publishes its inventory of unclassified and non-sensitive AI use cases under the Advancing American AI Act of December 2022, Executive Order 13960 of December 2020 on promoting the use of trustworthy artificial intelligence in the federal government, and Office of Management and Budget Memorandum M-25-21 of April 2025 on accelerating federal use of AI through innovation, governance and public trust (Source: techpolicy.press). The department also operates a search tool over the inventory at dhs.gov, which as of early 2026 was not current with the latest published inventory data (Source: techpolicy.press; dhs.gov).
The inventory released on January 28, 2026 recorded more than 200 AI use cases deployed or in development across DHS and its components, an increase of close to 40% over the July 2025 disclosure. ICE accounted for most of the growth, adding roughly two dozen applications including tools to process tips, review social media and mobile-device data, confirm identities using biometric data, and detect intentional misidentification (Source: fedscoop.com; techpolicy.press).
The inventory records only part of the department's AI use. The Brennan Center for Justice, analysing DHS surveillance contracting alongside the inventory, found that the inventory "discloses just a portion of the ways the department uses AI" (Source: brennancenter.org). A separate reading by the American Immigration Council argued that the mid-2025 update, which moved several programs to inactive status, reflected a reorganisation toward large vendor-built platforms rather than a reduction in AI use (Source: americanimmigrationcouncil.org).
AI uses and deployments
Immigration and Customs Enforcement (ICE)
ICE has stated that it uses AI tools from Palantir Technologies and OpenAI, as reported by The Information on January 29, 2026 ("ICE Says It Uses AI From Palantir, OpenAI"). The tools are applied to case management, intelligence analysis, and targeting.
Palantir appeared in the DHS AI inventory for the first time in the January 2026 edition, although it had long been reported as a DHS vendor; the inventory records use of the platform beginning in May 2025. Use of the platform was first reported by 404 Media and has since drawn scrutiny from Democratic lawmakers seeking to limit its application (Source: fedscoop.com).
Two ICE facial-recognition use cases are recorded in the inventory as deployed and as high-impact: Facial Recognition for Locating Vulnerable Populations (DHS-2457), under which Homeland Security Investigations submits facial photographs to an AI-enabled service that compares them against publicly available online imagery to generate leads on people such as those who entered the United States as unaccompanied minors; and Facial Recognition for National Security Investigations (DHS-2458), which applies the same technique to subjects, witnesses and associates in authorized national security investigations. DHS describes both as lead-generation tools that search public imagery rather than real-time surveillance systems, states that any candidate match must be corroborated by trained personnel, and records for each the potential impact that the service "may return too many candidates, resulting in the collection of irrelevant personal information," mitigated by a confidence-score threshold (Source: dhs.gov).
Beyond the inventory, ICE has been reported to rely on vendor-built platforms that combine data types in one place, including Palantir's immigration systems, Clearview AI facial recognition, BI2 iris-scanning devices and Paragon phone-forensics software (Source: americanimmigrationcouncil.org). Homeland Security Investigations entered a $9.2 million contract with Clearview AI for facial-image search across what the contract describes as a proprietary database of more than 50 billion facial images (Source: immpolicytracking.org). ICE has also obtained data from the nationwide licence-plate-reader network operated by Flock Safety by having other agencies run searches on its behalf (Source: brennancenter.org). The inventory additionally records administrative applications, including an AI-assisted résumé screening tool (Source: fedscoop.com).
Customs and Border Protection (CBP)
CBP entered a two-year contract with Altana AI on October 31, 2025 for AI-enabled trade-compliance screening, described as central to enforcing Section 232 national security tariffs on steel and aluminum. CBP also uses facial recognition at borders, drawing on systems tested through the Face Recognition Vendor Test (FRVT). Reporting by MIT Tech Review (August 14, 2024) described a proposal to use facial recognition to identify migrant children as they age ("The US wants to use facial recognition to identify migrant children as they age").
The inventory records further CBP applications: Integrated Defense and Security Solutions (DHS-311), computed-tomography x-ray systems with automated recognition for inline screening of high-volume parcels to detect contraband; the CBP Common AI Service image-analysis and data-correlation use case (DHS-2565), which is not live in production and carries an authority to test only within the platform; and Traffic Jam (DHS-2570), which applies facial recognition, AI and geospatial analysis to human-trafficking data (Source: dhs.gov). CBP has a $225,000 contract with Clearview AI covering facial-recognition technology and access to a database of more than 60 billion publicly available images; according to Wired, the statement of work embeds the technology in analysts' routine intelligence work rather than reserving it for investigations, and anticipates that analysts will handle biometric identifiers including facial images (Source: immpolicytracking.org).
Cybersecurity and Infrastructure Security Agency (CISA)
CISA's cyber tools are increasingly AI-augmented. The agency's handling of AI tools has also drawn oversight attention: Sen. Grassley Letter to CISA re ChatGPT (2026-02-05) concerns the acting director's uploading of sensitive documents to public ChatGPT.
Surveillance and facial recognition
DHS and CBP use of facial recognition at borders sits within a broader set of identity and surveillance deployments. Clear, an identity-management company, has expanded beyond airports (MIT Tech Review, November 20, 2024). MIT Tech Review reporting (May 12, 2025) described a new type of AI used by police to skirt facial recognition bans ("How a new type of AI is helping police skirt facial recognition bans"), while separate reporting tracked the movement to limit facial recognition (MIT Tech Review, July 20, 2023 onward).
At least 23 applications in the January 2026 inventory use some form of facial recognition, face matching or related biometric identification (Source: techpolicy.press). Among them is Mobile Fortify, a facial-recognition and fingerprint-matching application that compares biometric information against agency records and has been used by both CBP and ICE since May 2025. It appears twice in the inventory, indicating that CBP operates the system and ICE accesses it. 404 Media documented a case in which the application misidentified a woman twice during an immigration raid, notwithstanding ICE's characterisation of its results as a definitive determination of immigration status (Source: techpolicy.press; fedscoop.com).
DHS has taken down its published policy on the use of facial recognition, leaving the governing policy, if any, unclear (Source: brennancenter.org).
Media production
DHS has used Google and Adobe AI tools to produce videos, as reported by MIT Tech Review on January 29, 2026 ("DHS is using Google and Adobe AI to make videos").
Oversight and controversy
Several of the use cases added in the January 2026 inventory were classified as not high-impact, a designation that fell disproportionately to CBP and ICE entries and that experts questioned. For the Mobile Fortify application, deployed in immigration operations, ICE had not completed an AI impact assessment and no potential impacts had been identified in the inventory (Source: fedscoop.com).
The American Civil Liberties Union has sued over what it describes as a pattern of suspicionless stops, warrantless arrests and racial profiling by ICE and CBP in Minnesota, including through the use of facial recognition (Source: techpolicy.press). Reporting has described federal immigration officers equipped with facial recognition, biometric trackers, licence-plate readers, cell-phone location data, spyware and drones, including in the surveillance of protests (Source: immpolicytracking.org).
Open questions
- The share of DHS AI use that falls outside the published inventory is not established; the Brennan Center states only that the inventory is partial.
- What policy governs DHS facial-recognition use following the withdrawal of the published policy has not been disclosed.
Relationships
- deployed-by: Palantir Technologies, OpenAI, Clearview AI, Flock Safety, Altana AI, Clear, Google, Adobe.
- related: AI and Surveillance, Surveillance Technology, ICE — Immigration and Customs Enforcement (AI Deployer), CBP — Customs and Border Protection (AI Deployer), CISA — Cybersecurity and Infrastructure Security Agency (AI Deployer), American Civil Liberties Union (ACLU), Sen. Grassley Letter to CISA re ChatGPT (2026-02-05), Federal AI Adoption — Patterns and Tensions, Government AI Procurement, OMB Memorandum M-24-10.