The United States Immigration and Customs Enforcement (ICE) uses AI in its day-to-day activities to promote homeland security and public safety through the criminal and civil enforcement of federal laws governing border control, customs, trade, and immigration.
Below is an overview of each AI use case within ICE, as part of the Simplified DHS AI Use Case Inventory. More details about these use cases are available in the Full DHS AI Use Case Inventory on the DHS AI Use Case Inventory publication library.
AI use cases are listed by deployment status:
Pre-Deployment
Use Case Name: Machine Learning Translation Technology Initiative
Use Case ID: DHS-197
Use Case Summary: ICE’s Enforcement and Removal Operations (ERO) is developing a real-time translation tool to assist staff in communicating with noncitizens who have limited English proficiency. This tool, intended for non-critical conversations, will supplement existing language services contracts by providing real-time capabilities during various touch points in the ICE process. ERO personnel, who regularly interact with individuals who speak little or no English, currently rely on professional interpretation and translation services, which can be slow to secure. The new machine learning (ML) solution aims to facilitate informal communication quickly and efficiently. When deployed, it will not be used alone for materials vital to an individual’s rights or benefits, or when the source materials contain non-literal language, lack clear grammar, or are overly complex. In such cases, ERO personnel must seek additional language assistance via bilingual staff or professional language lines.
The initiative, first spearheaded by the U.S. Coast Guard for maritime usage with support from the DHS Science and Technology Directorate through the Silicon Valley Innovation Program (SVIP), received Fiscal Year 2022 funding for proof of concept. Solutions, in the form of mobile phone applications, are being developed under research and development phases and are projected to meet ERO’s requirements by January 2026. The Machine Learning Translation Technology Initiative will offer real-time communication and translation services, providing voice-to-text, text-to-voice, and voice-to-voice translations in at least 21 languages. This technology will enhance the ability of ICE ERO personnel to provide meaningful access to programs and services for limited English proficient (LEP) noncitizens.
Use Case Topic Area: Law & Justice
Deployment Status: Pre-deployment (Acquisition and/or Development)
Safety- and/or rights-impacting? No
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Title III Semantic Search and Summarization for Translated Content
Use Case ID: DHS-206
Use Case Summary: Homeland Security Investigations (HSI) is a critical component of ICE, responsible for investigating a wide range of crimes that threaten national security, public safety, and the economy. With a mission to protect the United States from transnational crime and terrorism, HSI works to disrupt and dismantle complex organizations that engage in human trafficking, narcotics smuggling, cybercrime, and other illicit activities. During an investigation, HSI collects large volumes of legally obtained evidentiary data, often in languages other than English. Once this data has been translated using translation and transcription services, agents and analysts need a better way to search through the vast amounts of data. The Title III Semantic Search and Summarization functionality will augment translation and transcription services by extracting relevant data using machine learning (ML) and Natural Language Processing (NLP) for correlation and semantic search. Results can then be summarized using a Large Language Model (LLM), giving users a tool to target relevant data only. This capability accelerates investigative analysis by rapidly identifying persons of interest, surfacing trends, and detecting networks or fraud, saving hundreds of hours in manual analysis. HSI will use this tool to generate leads, and further action will be required by investigators and analysts as part of the full investigative process before any action is taken against an individual.
Use Case Topic Area: Law & Justice
Deployment Status: Pre-deployment (Initiation)
Safety- and/or rights-impacting? No, use case is too new to fully assess impacts; will be reassessed before end of initiation stage.
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Policy Analyst Assistant
Use Case ID: DHS-208
Use Case Summary: The Student and Exchange Visitor Program (SEVP) uses AI to help policy analysts quickly find and summarize information about visa rules for students and schools, saving time and enabling faster responses to questions. SEVP is responsible for a wide variety of functions focusing on visa holders including F and M nonimmigrants and schools certified to admit them. The SEVP Regulation and Guidance Team performs research on large volumes of immigration policies, regulations, and other guidance for nonimmigrants to generate responses to data calls and queries from schools and students. Current methods for generating responses are manual and time-intensive. SEVP plans to leverage AI by developing a semantic search and LLM-based summarization tool using Retrieval Augmented Generation (RAG) to allow policy analysts to research and generate an initial analysis of applicable material more quickly and effectively. The generated output will be used as a starting point to assist policy analysts in completing their tasks, who will further refine, modify, vet, and review the response as part of their process. The source information provided to the tool is either in the public domain or is internal, non-sensitive data, with no Sensitive Personally Identifiable Information (SPII), Personally Identifiable Information (PII), nor Security Sensitive Information (SSI) used. This enhanced search capability will save hours of manual work per request, allowing analysts to respond more quickly with the most relevant information and focus on more complex policy and guidance issues.
Use Case Topic Area: Government Services (includes Benefits and Service Delivery)
Deployment Status: Pre-deployment (Acquisition and/or Development)
Safety- and/or rights-impacting? No
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: SEVP Response Center Chatbot – SID (SEVIS Interactive Dialog)
Use Case ID: DHS-2402
Use Case Summary: The Student and Exchange Visitor Program (SEVP) has a response center that answers questions about policy, procedures, and issues with Student and Exchange Visitor Information System (SEVIS). To provide faster answers, SEVP uses a chatbot named SEVIS Interactive Dialog (SID) to handle common questions from students and officials. If SID cannot help, it connects the caller to a human agent and shares the conversation to save time. Callers include visa holders (F, M, and J nonimmigrants), school and program officials, government users, and members of the public. Many questions are routine inquiries about established policies and procedures. SID is a conversational chatbot that uses voice recognition and Natural Language Understanding to understand and reply to users’ questions with scripted answers. SID is deterministic and does not use generative AI. It answers frequently asked questions but SEVP will not be programmed SID to answer safety- and/or rights-impacting questions. If SID cannot answer a question, it transfers the caller to an agent in the response center. The chatbot captures the interaction with the caller and sends the information via an API to Student and Exchange Visitor Program Automated Management System (SEVPAMS). SEVPAMS creates a ticket, either for calls completed by SID or those handed over to an agent. This allows agents to see the dialog between SID and the caller, reducing the need to repeat information. This frees up the human agents to handle more complex cases and specific record issues.
Use Case Topic Area: Government Services (includes Benefits and Service Delivery)
Deployment Status: Pre-deployment (Acquisition and/or Development)
Safety- and/or rights-impacting? No
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Digital Record Manager (DRM) User Assistance Chatbot
Use Case ID: DHS-2423
Use Case Summary: The Digital Records Manager (DRM) User Assistance Chatbot is an AI tool designed to help investigators efficiently search and gather information. Homeland Security Investigations (HSI), a component DHS, is responsible for investigating, disrupting, and dismantling transnational criminal organizations and terrorist networks that threaten or seek to exploit the customs and immigration laws of the United States. HSI uses the Investigative Case Management (ICM) system to document its investigative activities and has integrated the DRM with ICM to manage digital media associated with investigations. Together, ICM and DRM serve as HSI’s official system of record for investigative case data, ensuring compliance with the National Archives and Records Administration’s mandate for fully electronic records. The HSI DRM User Assistance Chatbot provides users with a consistent location in the DRM interface to pose natural language questions about using the application for case and media management. It uses a Retrieval Augmented Generation (RAG) approach to augment natural language responses from a Large Language Model (LLM) with a custom Knowledge Base containing DRM documentation. The Chatbot outputs natural language responses to user questions, guiding them to accomplish specific tasks within the system. This immediate, on-demand assistance emulates help desk support, reducing time-to-completion for many DRM functions and easing the burden on help desk staff.
Use Case Topic Area: Mission-Enabling (internal agency support)
Deployment Status: Pre-deployment (Initiation)
Safety- and/or rights-impacting? No
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Burlington Finance Center Voice Bot
Use Case ID: DHS-2436
Use Case Summary: The Bond Management Information System (BMIS) supports the lifecycle of an immigration bond, from posting to disposal. The Burlington Finance Center (BFC) Voice Bot will be a public-facing multilingual chatbot that uses voice recognition to understand and respond to suer inquiries over the phone. It employees Natural Language Understanding (NLU) and Natural Language Processing (NLP) for voice-to-text and text-to-voice translations, enabling it to recognize voices and interpret meaning. The BFC Voice Bot will be deterministic and will not use Generative AI. The BFC Voice Bot will identify and verify callers, providing or retrieving and sharing bond status information. It will support English, Spanish, and French/Haitian Creole. The current bond inquiry workflow is entirely manual; automating the process will improve efficiency and reduce the burden of the Bonds Management team.
Use Case Topic Area: Government Services (includes Benefits and Service Delivery)
Deployment Status: Pre-deployment (Initiation)
Safety- and/or rights-impacting? No, use case is too new to fully assess impacts; will be reassessed before end of initiation stage.
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Email Analytics for Investigative Data (formerly Email Analytics)
Use Case ID: DHS-48
Use Case Summary: Homeland Security Investigations (HSI) is a critical component ICE, responsible for investigating a wide range of crimes that threaten national security, public safety, and the economy. HSI targets transnational crime and terrorism by disrupting organizations involved in human trafficking, narcotics smuggling, cybercrime, and other illicit activities. To achieve this mission, HSI uses on advanced technologies to analyze and process vast amounts of data, including audio and video evidence. HSI personnel handle significant amounts of legally acquired, multilingual email data, which must be prepared (ingested, triaged, translated, searched and filtered) before analysis. The Email Analytics application automates this process, using machine learning (ML) for spam classification, translation, and entity extraction (such as names, organizations, or locations). The Translation and Transcription Tool translates emails into English, reducing the time and resources needed for data preparation. This increases the analytic utility of the data, allowing HSI personnel to conduct quicker analysis. HSI uses this tool to generate leads, with further actions taken by investigators and analysts as part of the full investigative process before any action is taken against an individual.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? No. It was presumed rights-impacting relating to law enforcement in certain contexts and language translation in official contexts, but DHS Chief AI Officer determined this use case does not satisfy the definition of rights-impacting AI in M-24-10. This use case provides more efficient data processing for HSI personnel. The AI output may be used to produce investigative insights in the form of data, information leads or connections that HSI personnel can use to inform investigations, but the output itself is data preparation and organization so HSI personnel can produce those leads when combining the AI output with the personnel’s expertise and other relevant investigative data and information. Personnel may use these insights for law enforcement purposes in ongoing investigations with existing targets to assist in activities such as producing risk assessments about individuals or identifying criminal suspects; however, all insights are reviewed and validated by both personnel and supervisors before being included in any official case management system, and do not serve as a principal basis for law enforcement action or decision. Read more about safety and/or rights-impacting AI.
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Mobile Device Analytics for Investigative Data (formerly Mobile Device Analytics)
Use Case ID: DHS-49
Use Case Summary: Homeland Security Investigations (HSI) is a critical component of ICE, responsible for investigating a wide range of crimes that threaten national security, public safety, and the economy. With a mission to protect the United States from transnational crime and terrorism, HSI works to disrupt and dismantle complex organizations that engage in human trafficking, narcotics smuggling, cybercrime, and other illicit activities. To achieve this mission, HSI relies on advanced technologies and innovative solutions to analyze and process vast amounts of data, including audio and video evidence. Mobile device analytics capabilities were developed to meet the growing need for investigators to quickly and effectively review and analyze the massive amounts of data extracted from mobile devices obtained in court-ordered seizures through a warrant or other investigative due process. The solution empowers investigators and analysts to identify and extract critical evidence, relationships, and networks from mobile device data, leveraging machine learning capabilities to determine locations of interest. Present capabilities are limited to the use of an algorithm for clustering of geo-location data, allowing investigators to identify 'stops' and 'overlaps' of locations from lawfully seized phones, where a 'stop' is a location a phone may have remained for a period and an 'overlap' is a location where multiple phones may have visited. By streamlining the analysis process, these capabilities aim to enhance the efficacy of HSI agents and analysts by providing advanced data processing to identify the most relevant information for lead generation in a shorter amount of time than manually processing, ultimately contributing to the disruption and dismantling of criminal networks and the protection of national security. HSI uses this tool to generate leads, and further action is required by investigators and analysts as part of the full investigative process before any action is taken against an individual.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? No. It was presumed rights-impacting relating to law enforcement in certain contexts and language translation in official contexts, but the DHS Chief AI Officer determined this use case does not satisfy the definition of rights-impacting AI in M-24-10. This use case provides more efficient data processing for geo-location data extracted from mobile devices obtained in court-ordered seizures through a warrant or other investigative due process. An ML algorithm is used to identify locations where a phone stopped for a period of time, as well as locations where multiple seized phones may have stopped, reducing the time it would take to gain the same insights from manual analysis. This output may be used to produce investigative insights in the form of data, information leads or connections that HSI personnel can use to inform investigations, but the output itself is data preparation and organization so HSI personnel can produce those leads when combining the AI output with the personnel’s expertise and other relevant investigative data and information. All data insights are reviewed and validated by both personnel and supervisors before being included in any official case management system, and do not serve as a principal basis for law enforcement action or decision. Read more about safety and/or rights-impacting AI.
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Identification Card and Travel Document Code Detection (formerly Barcode Scanner)
Use Case ID: DHS-53
Use Case Summary: Homeland Security Investigations (HSI) is a critical component of ICE, responsible for investigating crimes that threaten national security, public safety, and the economy. HSI’s mission is to protect the United States from transnational crime and terrorism by disrupting organizations involved in human trafficking, narcotics smuggling, cybercrime, and other illicit activities. To achieve this mission, HSI uses advanced technologies to analyze and process vast amounts of data, including audio and video evidence. The Identification Card and Travel Document Code Detection service leverages mobile phone cameras to read 2D Data Matrix Codes on U.S. driver's licenses and identification cards, as well as Machine Readable Zones (MRZ) on travel documents like passports and passport cards. This service uses machine learning (ML) to enhance code detection, allowing for seamless scanning and automatic population of detected information into corresponding text fields within HSI mobile apps. By streamlining the code recognition process, this service reduces the time required for users to read cards or documents, eliminating the need for manual data entry, and resulting in more efficient in-person interactions with individuals.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? No
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Normalization Services
Use Case ID: DHS-P1
Use Case Summary: Homeland Security Investigations (HSI) is a critical component of ICE, responsible for investigating a wide range of crimes that threaten national security, public safety, and the economy. With a mission to protect the United States from transnational crime and terrorism, HSI works to disrupt and dismantle complex organizations that engage in human trafficking, narcotics smuggling, cybercrime, and other illicit activities. To achieve this mission, HSI relies on advanced technologies and innovative solutions to analyze and process vast amounts of data, including audio and video evidence.
HSI utilizes machine learning to enhance data accuracy and efficiency by verifying, validating, correcting, and normalizing various types of information, including addresses, phone numbers, names, and identification numbers. This process helps to eliminate data entry errors, detect intentional misidentification, and connect related information across multiple datasets, ultimately reducing the time and resources required for investigations. The machine learning-powered normalization services offered by HSI include converting ambiguous addresses into usable formats, identifying identification types from partial information, categorizing names with complex suffixes and family names, and standardizing phone numbers to the E164 format, including determining their originating county. By normalizing and improving the quality of investigative datasets, HSI is able to use more advanced tools to find correlations and leads that would have otherwise gone undetected without extensive manual effort.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? No
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Voice Analytics for Investigative Data
Use Case ID: DHS-123
Use Case Summary: Homeland Security Investigations (HSI) is a critical component of ICE, responsible for investigating crimes that threaten national security, public safety, and the economy. HSI’s mission is to protect the United States from transnational crime and terrorism by disrupting organizations involved in human trafficking, narcotics smuggling, cybercrime, and other illicit activities. During investigations, HSI encounters significant volumes of live and recorded content that require language extraction and analysis. Currently, this is done manually, which is labor-intensive and costly. To address this challenge, HSI has developed Voice Analytics, a cutting-edge technology that uses machine learning (ML) and advanced speech processing techniques for multilingual speech transcription and translation of audio files. This system outputs transcribed and translated audio files, enabling HSI personnel to quickly identify new investigative leads and analyze the outputs against other law enforcement datasets, significantly improving the efficiency and effectiveness of their investigations compared to manual processes. HSI uses this tool to generate leads, with further actions taken by investigators and analysts as part of the full investigative process before any action is taken against an individual.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Implementation and Assessment)
Safety- and/or rights-impacting? No. It was presumed rights-impacting relating to law enforcement in certain contexts and language translation in official contexts, but the DHS Chief AI Officer determined this use case does not satisfy the definition of rights-impacting AI in M-24-10. This use case provides more efficient data translation and data management for HSI investigations. The AI output may be used to produce investigative insights, but the output itself is supporting information for HSI personnel to use to produce leads. Personnel may use these insights for law enforcement purposes in ongoing investigations with existing targets to assist in activities such as producing risk assessments about individuals or identifying criminal suspects; however, all insights are reviewed and validated by both personnel and supervisors before being included in any official case management system, and do not serve as a principal basis for any law enforcement action or decision. Read more about safety and/or rights-impacting AI.
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Investigative Prioritization Aggregator
Use Case ID: DHS-125
Use Case Summary: Homeland Security Investigations (HSI) is a critical component of ICE, responsible for investigating crimes that threaten national security, public safety, and the economy. HSI’s mission is to protect the United States from transnational crime and terrorism by disrupting organizations involved in human trafficking, narcotics smuggling, cybercrime, and other illicit activities. HSI relies on advanced technologies and innovative solutions to analyze and process vast amounts of data, including audio and video evidence. The sheer volume of data often overwhelms human capabilities, making it challenging for HSI personnel to analyze evidence and identify key players in criminal networks. Currently, there is no effective mechanism to quantify the level of evidence related to a particular subject or entity, or to determine the most influential actors within a network. This is particularly critical in the counter-opioid/fentanyl mission, where timely and accurate intelligence is essential. To address this challenge, HSI has developed a project that uses machine learning (ML) to assign point values to data, enabling the scoring of information associated with a given selector, such as a phone number or legal name. This scoring system helps to understand the importance of an entity to investigations and the potential consequences of removing or neutralizing that entity. By doing so, HSI personnel can focus on high-priority targets and associated criminal networks, enhancing their ability to disrupt and dismantle these threats. HSI uses this tool to generate leads, with further actions taken by investigators and analysts as part of the full investigative process before any action is taken against an individual.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? No. It was presumed rights-impacting relating to law enforcement in certain contexts, but the DHS Chief AI Officer determined this use case does not satisfy the definition of rights-impacting AI in M-24-10. This use case provides more efficient data analysis for HSI investigations. The AI outputs may be used to produce investigative insights, such as data correlations, information leads, or connections that HSI personnel can use to inform investigations. The output itself is supporting information (e.g. scoring/ranking importance of data entities) to help HSI personnel focus on the most relevant information during an investigation. Personnel may use these insights for law enforcement purposes in ongoing investigations with existing targets to assist in activities such as producing risk assessments about individuals or identifying criminal suspects; however, all insights are reviewed and validated by both personnel and supervisors before being included in any official case management system, and do not serve as a principal basis for any law enforcement action or decision. Read more about safety and/or rights-impacting AI.
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Video Analysis Tool
Use Case ID: DHS-172
Use Case Summary: Homeland Security Investigations (HSI) is a critical component of ICE, responsible for investigating crimes that threaten national security, public safety, and the economy. HSI’s mission is to protect the United States from transnational crime and terrorism by disrupting organizations involved in human trafficking, narcotics smuggling, cybercrime, and other illicit activities. HSI relies on advanced technologies to analyze and process vast amounts of data, including audio and video evidence. The Video Analysis Tool (VAT) is used to investigate human rights violations, detect fraud, and counter transnational organized crime involved in synthetic opioids. Machine learning (ML) algorithms identify human faces and crop the image from media. VAT leverages model-based matching and developed algorithms that are indifferent to subject pose, illumination, and expression. The output is a collection of relevant facial images of perpetrators, which are used to create a database of suspects. These images are then used to query and compare with relevant Federal biometric and biographical databases through connections to VAT, or shared with other agency partners, where the images are compared against those partners’ holdings using facial matching methodologies. HSI uses this tool to generate leads, with further action taken by investigators and analysts as part of the full investigative process before any action is taken against an individual.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? Yes. Rights-impacting
Key Identified Risks & Mitigations: Forthcoming. Office of Management and Budget (OMB) approved a compliance extension through November 30, 2025. The requested extension is to provide time to get additional foundational information from the vendor, such as information about data provenance and training, model accuracy, and ongoing monitoring of model's performance.
Read more about safety and/or rights-impacting AI and compliance with required minimum risk management practices.
Face Recognition/Face Capture (FR/FC)? Yes. All Face Recognition and Face Capture (FR/FC) technology is tested both prior to operational use and at least every three years during operational use. DHS Science and Technology (S&T) oversees testing and evaluation based on International Organization for Standardization/ International Electrotechnical Commission (ISO/IEC) standards and technical guidance issued by National Institute of Standards and Technology (NIST). DHS S&T applies laboratory, scenario, and operational testing to cost-effectively characterize technology performance and, when feasible, disaggregate performance by user demographics such as gender, age, and skin tone.
Use Case Name: Semantic Search and Summarization for Investigative Reports (formerly Semantic Search and Summarization)
Use Case ID: DHS-204
Use Case Summary: Homeland Security Investigations (HSI) is a critical component of ICE, responsible for investigating crimes that threaten national security, public safety, and the economy. HSI’s mission is to protect the United States from transnational crime and terrorism by disrupting organizations involved in human trafficking, narcotics smuggling, cybercrime, and other illicit activities. HSI relies on advanced technologies to analyze and process vast amounts of data, including audio and video evidence. Effective investigations depend on the ability to extract relevant information quickly and accurately from large volumes of unstructured reports. Traditional lexical search methods often fall short in this regard. Semantic Search and Summarization technology bridges this gap by leveraging large language models (LLM) to allow investigators to submit natural language queries. These queries are analyzed to identify the most relevant information in reports, accelerating investigative analysis by rapidly identifying persons of interest, surfacing trends, and detecting networks or fraud. Furthermore, the system allows investigators to build reports by entering raw notes and leveraging LLM to generate a draft report that adheres to the Report of Investigation (ROI) manual's formatting and style guidelines. This saves time and produces more articulate and accurate reports. Once reviewed, validated, and approved by supervisors, the final report is uploaded to a case management system. HSI uses this tool to generate leads, with further action required by investigators and analysts as part of the full investigative process before any action is taken against an individual.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Implementation and Assessment)
Safety- and/or rights-impacting? No. It was presumed rights-impacting relating to law enforcement in certain contexts and language translation in official contexts, but the DHS Chief AI Officer determined this use case does not satisfy the definition of rights-impacting AI in M-24-10. This use case provides a natural language contextual search and summarization capability against existing Reports of Investigation, producing more relevant search responses and investigative insights in the form of data, information leads or connections that HSI personnel can use to inform investigations. The AI output (search queries and responses) are not used as a principal basis for any action but provide investigators with easy to read search responses that are accompanied by links to source material for further analysis. Any data used to produce these investigative insights are first obtained through legal means and processes for the purposes of law enforcement investigations and do not significantly impact the categories of rights listed in the definitions of “rights-impacting AI.” Personnel may use these insights for law enforcement purposes in ongoing investigations with existing targets to assist in activities such as producing risk assessments about individuals or identifying criminal suspects; however, all insights are reviewed and validated by both personnel and supervisors before being included in any official case management system, and do not serve as a principal basis for enforcement decisions. Read more about safety and/or rights-impacting AI.
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Facial Recognition for Investigations of Child Sexual Exploitation and Abuse
Use Case ID: DHS-362
Use Case Summary: The Homeland Security Investigations (HSI) Child Exploitation Investigations Unit (CEIU) handles cases involving the production, distribution, and possession of child sexual exploitation materials. It works to identify and rescue victims using advanced forensic technology and investigative techniques. CEIU employs an “image recognition neural network” to help identify unknown victims and offenders depicted in child sexual abuse material (CSAM). HSI personnel receive specific training to use this technology, examining newly discovered and unidentified images of CSAM. The technology generates leads on the possible identities of victims and offenders, but no enforcement action is taken based on these leads alone. Potential identifications must be further investigated and validated. CEIU ensures the tool’s use complies with all policy and privacy standards as set forth in the Privacy Impact Assessment. This technology has been proven effective in identifying and rescuing children who might not have been found through traditional methods and in preventing further victimization by arresting otherwise undetected offenders. The tool provides visually similar photos of individuals, serving as a starting point for further investigation by analysts. No enforcement action is permitted on these leads alone.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? Yes. Rights-impacting
Key Identified Risks & Mitigations: The AI-enabled facial recognition service may return excessive amounts of candidates, leading to an over-collection of individual information irrelevant to the ongoing criminal case. The AI-enabled facial recognition service only returns candidates that meet or exceed a confidence score threshold. Results are ranked so that candidates with the highest confidence score are returned first. Only successfully vetted candidates will be entered into ICE Report of Investigation.
Read more about safety and/or rights-impacting AI and compliance with required minimum risk management practices.
Face Recognition/Face Capture (FR/FC)? Yes. All Face Recognition and Face Capture (FR/FC) technology is tested both prior to operational use and at least every three years during operational use. DHS Science and Technology (S&T) oversees testing and evaluation based on International Organization for Standardization/ International Electrotechnical Commission (ISO/IEC) standards and technical guidance issued by National Institute of Standards and Technology (NIST). DHS S&T applies laboratory, scenario, and operational testing to cost-effectively characterize technology performance and, when feasible, disaggregate performance by user demographics such as gender, age, and skin tone.
The DHS 2024 report on Use of Face Recognition and Face Capture Technologies provides additional information about use of FR/FC at DHS, including testing and evaluation of this use case.
Use Case Name: Biometric Check-in for ATD-ISAP (SmartLINK)
Use Case ID: DHS-407
Use Case Summary: The Intensive Supervision Appearance Program (ISAP) Monitoring App includes unique facial verification technology that allows participants to capture their photo during ICE Alternatives to Detention (ATD) enrollment for future verification purposes (e.g., during check-in). Unlike facial identification or recognition, the facial verification function uses a one-to-one matching approach of the individual presenting to the Monitoring App during their check-in by comparing it against the ATD participant’s photograph captured upon enrollment to determine whether the person is who they declare themselves to be. The Monitoring App can be launched from either an ICE ATD-issued mobile device or from the participant’s personal mobile device. It takes a series of photos of the participant during enrollment, which are stored in the ISAP ATD case management system. During subsequently checks in, an automatic 1:1 verification image of the individual in front of the phone’s camera is matched against the stored profile images to enable the ICE ATD Case Manager to verify the participant’s identity, using a proprietary algorithm. The technology also recognizes if a “live” person is in front of the camera, as opposed to a representation or image of a person. There is a visual ICE ATD Case Manager review if there is no match or if the matching attempt generates an error alert. The photos are only used for the described verification purpose and are not shared for any other purpose or with any other entity or database. Any pictures taken during the facial template capture are immediately deleted from the device and only the facial measurements are captured.
The ISAP Biometric Monitoring App is a technology option that allows participants to report in using a smartphone. This app verifies a participant’s identity, determines their location, and quickly collects status change information. It adds functionality not available with telephonic methods and is less intrusive than a global positioning system unit. The ISAP Monitoring App limits in-person interactions for routine check-ins, allowing more time to be allocated to non-compliant participants, complex removal proceedings cases, and docket management.
There are two outputs related to using ISAP Biometric Monitoring App: either a participant passes the biometric match, or the photo is moved to a pending review status.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? Yes. Rights-impacting
Key Identified Risks & Mitigations: Environmental variables may negatively impact model performance. At enrollment, participants meet with a Case Manager, who explains the program in the participant’s preferred language. Participants receive guidance on the use of the system and environmental parameters which may affect results (e.g. poor lighting, busy backgrounds). All nonmatches are reviewed by human examiner and considered for a redo before the check in is officially marked as a nonmatch.
Read more about safety and/or rights-impacting AI and compliance with required minimum risk management practices.
Face Recognition/Face Capture (FR/FC)? Yes. All Face Recognition and Face Capture (FR/FC) technology is tested both prior to operational use and at least every three years during operational use. DHS Science and Technology (S&T) oversees testing and evaluation based on International Organization for Standardization/ International Electrotechnical Commission (ISO/IEC) standards and technical guidance issued by National Institute of Standards and Technology (NIST). DHS S&T applies laboratory, scenario, and operational testing to cost-effectively characterize technology performance and, when feasible, disaggregate performance by user demographics such as gender, age, and skin tone.
Use Case Name: Virtual Agent Chatbot for Human Capital
Use Case ID: DHS-2307
Use Case Summary: ICE’s Human Capital Virtual Agent “Alice” is a chatbot available to all ICE employees focused on delivering answers and resources for Human Capital related inquiries through conversation. The chatbot will be enhanced through the implementation of Natural Language Understanding (NLU) within ServiceNow, a type of AI focused on understanding and processing human language to enable effective interactions between the chatbot and users. Understanding natural language will enhance user experience by providing a way for the chatbot to comprehend what the employee is asking for and provide relevant and accurate responses, making it easier for users to get the help they need. By providing ICE employees information fast and efficiently at any time through Virtual Agent NLU, the chatbot aims to reduce the amount of time spent by Human Capital Specialists responding to the inquiries.
Use Case Topic Area: Education & Workforce
Deployment Status: (Implementation and Assessment)
Safety- and/or rights-impacting? No
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Hurricane Score
Use Case ID: DHS-2408
Use Case Summary: ICE Enforcement Removal Operations’ Alternatives to Detention (ATD) program exists to ensure compliance with release conditions and provides important case management services for non-detained noncitizens. ATD consists of the Intensive Supervision Appearance Program (ISAP), which utilizes case management and technology tools to support noncitizens compliance with release conditions, court hearings, and final orders of removal, while allowing them to remain in their communities as they move through the immigration process or prepare for departure from the United States. Once individuals are in the ATD-ISAP program, officers periodically perform case reviews to determine if current levels of case management and technology assignment are appropriate for the noncitizen in the program or if they need to be adjusted. During the case review, the officer will consider numerous factors to include, but not limited to, current immigration status, supervision and compliance history, pending benefits, being a care giver or care provider, current immigration stage, pending criminal history and/or convictions, the hurricane score, and other factors. The factor known as the Hurricane Score is a quasi-binomial, binary classification machine learning (ML) model that is given information by an analyst that is known about an individual (factors from case management details and participant actions) and determines the probability that the individual will abscond based on absconding patterns the model has learned from inactive ATD-ISAP case data. The model returns a score from 1-5, with the higher number equating to a higher risk. The officer may use this score as one of the many factors previously described when deciding the appropriate level of case management and technology for noncitizens enrolled in the ATD-ISAP program. Because the hurricane score can quickly evaluate an enormous amount of information on thousands of noncitizens in the ATD-ISAP program, it provides officers with additional insight that they would not have otherwise had when performing a case review.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? Yes. Rights-impacting
Key Identified Risks & Mitigations: There is a risk that the Absconder Model may include false positives and model bias. A false positive would occur when non-absconders are incorrectly classified with a score indicating a high probability of absconding. Model bias can arise if the training data is not representative of the full range of participants, leading to poor performance on certain patterns or characteristics. These risks are identified and if necessary remediated through error analysis during model testing and model performance monitoring.
Read more about safety and/or rights-impacting AI and compliance with required minimum risk management practices.
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: ICE Mobile Check-In Application
Use Case ID: DHS-2409
Use Case Summary: ICE is authorized under the Immigration and Nationality Act (INA) to set terms of supervision on non-citizens who are in immigration proceedings or who have completed proceedings and were not granted a status or relief of any kind. The ICE Check-In application is a mobile application where low-risk adult noncitizens can check-in using location and photo capabilities on user owned mobile devices (iOS and Android application stores). The Check-In App is an opt-in capability that allows for low-risk adult noncitizens to securely check-in via an ICE Verified user account, verify their identity, identify their location, and satisfy check-in requirements thereby reducing in-person office visits and reducing manual processing work. The app requires users to follow several randomly prompted actions and leverages programmatic logic-based heuristics to infer liveness based upon facial-tracking feedback from user interaction. AI and machine learning (ML) are used for this face detection and tracking. Once captured, the photo is verified with a 1:1 verification using the Customs and Border Protection (CBP) Traveler Verification Service. Verification of all requirements will satisfy check-in requirements. A failed verification will result in a review by an officer. The officer can request the user try again or request the user use the default in-person check-in process by making an appointment with their local ICE office. AI-enabled capabilities within the ICE Check-In application give ICE the ability to scale its interactions to a greater number of noncitizens and reduce in-office presence at ICE facilities.
Use Case Topic Area: Law & Justice
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? Yes. Rights-impacting
Key Identified Risks & Mitigations: There is a risk that the face detection model testing will show the model performs poorly for certain performance metrics that cannot be remediated through software configuration or other reasonable means. In that scenario, ICE will either pursue using a different face detection model or pursue additional testing to uncover the root cause of poor performance.
Read more about safety and/or rights-impacting AI and compliance with required minimum risk management practices.
Face Recognition/Face Capture (FR/FC)? Yes. All Face Recognition and Face Capture (FR/FC) technology is tested both prior to operational use and at least every three years during operational use. DHS Science and Technology (S&T) oversees testing and evaluation based on International Organization for Standardization/ International Electrotechnical Commission (ISO/IEC) standards and technical guidance issued by National Institute of Standards and Technology (NIST). DHS S&T applies laboratory, scenario, and operational testing to cost-effectively characterize technology performance and, when feasible, disaggregate performance by user demographics such as gender, age, and skin tone.
Use Case Name: AI Assisted Compromise Email Detector (AACED)
Use Case ID: DHS-2424
Use Case Summary: The AI Assisted Compromise Email Detector (AACED) helps prioritize which emails human security analysts should investigate first to identify cyber compromise ICE. This use case reports AI-enabled analysis on a one-time export of emails between ICE and Microsoft relevant to Cybersecurity and Infrastructure Security Agency (CISA) Emergency Directive 24-02. This use case does not perform live email analysis and only includes emails sent to/from ICE email accounts.
The Information Assurance Division (IAD) within ICE is responsible for providing Cybersecurity Services to ICE, including operating the ICE Security Operations and Computer Security Incident Response Center through its Cyber Defense & Intelligence Branch. The Branch performs information assurance mitigation, responds to (and reports on) Information Technology (IT) security incidents, performs systems vulnerability testing, and monitors and evaluates the security performance on sensitive enterprise systems. CISA published Emergency Directive (ED) 24-02: Mitigating the Significant Risk from Nation-State Compromise of Microsoft Corporate Email System in which all Federal Government Agencies were required to review the stored email metadata from Microsoft to determine if they were affected from the breach. This requires the investigation of tens of thousands of emails to determine if there were any signs of compromise or Personally Identifiable Information (PII) leakage.
The use case was developed to assist ICE Security Operations Center (SOC) in reviewing a collection of emails between ICE personnel and Microsoft that were part of ED 24-02. It provides a faster mechanism for SOC analysts to determine indicators of compromise, reducing the level of effort for these individuals’ analysis exponentially. To assist the analysts, Named Entity Recognition (NER) was used to detect PII and other associated keywords while a locally running Large Language Model (LLM), with no external connections, answers questions about selected emails. The LLM also answers preset questions formulated from Standard Operating Procedures (SOP). The analysts use the outputs to expedite the detection of PII and key language within emails, reducing the time needed to analyze the total collection of emails. If the use case detects PII or key language, then those emails are ranked as higher risk and prioritized by the analyst. The analyst then inspects the email at that location for presence of the data and logs it manually allowing for specified actions based on the results of the analysts’ review. Ultimately, all emails are manually inspected and depending on the circumstance of the investigation the ICE Privacy Office may be contacted notifying them of the PII that was exposed, or the analyst will provide the information to the SOC Leadership to initiate the incident response process to perform remediation actions needed. The use case only provides ranking for prioritized processing to the SOC analysts.
Use Case Topic Area: Mission-Enabling (internal agency support)
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? No
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Intelligent Document Processing for Workflow Automation
Use Case ID: DHS-2425
Use Case Summary: ICE processes forms of all types across its Program Offices, requiring a significant amount of time to manually validate and extract form data and enter it into a variety of systems. ICE uses a Commercial Off-The-Shelf (COTS) platform and cloud services to provide Robotic Process Automation (RPA) and AI-based intelligent document processing capabilities to the enterprise. Business units within ICE leverage these services to automate repeatable, time-consuming processes such as invoice processing, and form entry validation and extraction. This platform will provide Optical Character Recognition and machine learning models to verify, extract, and classify information from ICE forms such as contract invoices, the Disability Accommodation Notification (DAN) form, Notices to Appear (NTA), email inquiries to the Office of the Inspector General, the Sexual Abuse or Assault Incident Review form, Student and Exchange Visitor Program (SEVP) system access requests, the Notice to Student or Exchange Visitor (I-515) form, and the Training Plan for STEM OPT Students (I-983) form. AI is used during these automations to verify and extract information from the forms only and does not change data or make decisions based on the data. Additional workflow automations will use this data for follow-on actions such as processing an invoice, creating a case, or entering data into another system, depending on the purpose of the automation. The platform also includes the ability to create a feedback loop to continuously improve the model's accuracy by retraining it with new, relevant data. Using AI to provide information extraction for these processes saves ICE personnel a significant amount of time while improving data quality and enabling automation.
Use Case Topic Area: Mission-Enabling (internal agency support)
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? No
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Cybersecurity Threat Management, Detection, and Response
Use Case ID: DHS-2426
Use Case Summary: The Information Assurance Division (IAD) within ICE is responsible for providing internal Cybersecurity Services to ICE. Within IAD is the Cyber Defense & Intelligence (CD&I) Branch whose mission is operating the ICE Security Operations and Computer Security Incident Response Center. The Branch performs information assurance mitigation, responds to and reports on Information Technology (IT) security incidents, performs systems vulnerability testing, and monitors and evaluates the security performance on sensitive enterprise systems. The CD&I Branch performs monitoring and analysis on aggregate ICE cybersecurity data to identify and respond to anomalous and malicious activity. This data is limited to IT security operations and is not used to conduct AI-enabled workplace surveillance or automated personnel management. CD&I uses several AI-enabled cybersecurity tools to analyze this data. Machine learning (ML) models, such as classification and regression models, are used to analyze historical data and detect emerging threats through pattern recognition. Additional capabilities include the identification of real-time threats by using algorithms and ML from a vast database of known threats and patterns during continuous monitoring of ICE cybersecurity data. This includes recognizing phishing patterns, malware signatures, or abnormal network traffic patterns across a variety of tools. This provides security analysts with modern tools to identify and respond to threats much more quickly than previously possible, minimizing potential damage to systems and data.
Use Case Topic Area: Mission-Enabling (internal agency support)
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? No
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Translation and Transcription for Investigative Data
Use Case ID: DHS-2427
Use Case Summary: Homeland Security Investigations (HSI) is a critical component of ICE, responsible for investigating a wide range of crimes that threaten national security, public safety, and the economy. With a mission to protect the United States from transnational crime and terrorism, HSI works to disrupt and dismantle complex organizations that engage in human trafficking, narcotics smuggling, cybercrime, and other illicit activities. To achieve this mission, HSI relies on advanced technologies and innovative solutions to analyze and process vast amounts of data, including audio and video evidence. HSI investigators often encounter data from various sources, including legal and administrative processes, enforcement actions, and open-source materials, in languages other than English. To unlock the value of this data, it must be translated into English before further analysis can be conducted. The Translation and Transcription Service leverages neural machine translation (NMT) models for text translation and automatic speech recognition (ASR) and deep neural network (DNN) models with normalization for voice-to-text transcription. This innovative approach enables users to quickly triage large datasets and identify key information relevant to investigations. Any data deemed critical for court proceedings is then submitted to certified human translators for final review, ensuring that government resources are allocated efficiently and only used for necessary translations and transcriptions. HSI uses this tool to generate leads, and further action is required by investigators and analysts as part of the full investigative process before any action is taken against an individual.
Use Case Topic Area: Law & Justice, Science and Space
Deployment Status: Deployed (Operation and Maintenance)
Safety- and/or rights-impacting? No. It was presumed rights-impacting relating to law enforcement in certain contexts and language translation in official contexts, but the DHS Chief AI Officer determined this use case does not satisfy the definition of rights-impacting AI in M-24-10. This use case provides more efficient translation capabilities for non-English language data encountered during HSI investigations. The AI output may be used to produce investigative insights, but the output itself is supporting information for HSI personnel to use for ease of review and for further analysis. The translation being performed is not live translation and is not used for official communication to an individual, but rather for analysis of previously recorded audio file data that has been legally obtained during an investigation. Personnel may use these insights for law enforcement purposes in ongoing investigations with existing targets to assist in activities such as identifying criminal suspects; however, the translated data primarily allows investigators to identify the most relevant data that, if deemed critical for court proceedings, can be submitted to certified human translators for final review. This ensures that government resources are allocated efficiently and only used for necessary translations and transcriptions of relevant data. Read more about safety and/or rights-impacting AI.
Face Recognition/Face Capture (FR/FC)? No
Use Case Name: Machine Translation (Previously Language Translator)
Use Case ID: DHS-9
Use Case Summary: The Rapid Analysis and Visualization Engine for Reporting (RAVEn) platform provides capabilities to upload data that may contain text in languages other than English. To support the agent/analyst analysis of the ingested data as well as to provide for the means to standardize searches across RAVEn data sources, RAVEn provides language translation capabilities to create text conversions between the original language and English. Data translation can be performed on ingest or on an ad hoc basis. Solution will provide translation support of, at minimum, plain text, word documents, and portable document formats using Natural Language Processing and machine learning.
Deployment Status: Inactive (consolidated with another use case). This use case was consolidated under Translation and Transcription for Investigative Data (DHS-2427).
Use Case Name: Facial Recognition Service
Use Case ID: DHS-54
Use Case Summary: The Facial Recognition Service is used during investigations conducted by Homeland Security Investigations (HSI) agents and analysts for identification of known individuals, as well as extracting faces for further investigations from perpetrators including child exploitation offenses, human rights atrocities, and war criminals. This is a DHS HSI Innovation Lab/RAVEn project. The Repository for Analytics in a Virtualized Environment (RAVEn) facilitates large, complex analytical projects to support ICE’s mission to enforce and investigate violations of U.S. criminal, civil, and administrative laws. RAVEn also enables tools used to analyze trends and isolate criminal patterns as HSI mission needs arise. For more information, please read the DHS/ICE/PIA-055 - Privacy Impact Assessment 055 for RAVEn.
Deployment Status: Inactive (no longer used). This use case covered all face recognition use, which have been separated into individual, inventoried use cases such as Video Analysis Tool (VAT) (DHS-172).
Use Case Name: RAVEn Compliance Automation Tool (CAT)
Use Case ID: DHS-175
Use Case Summary: Repository for Analytics in a Virtualized Environment (RAVEn) Compliance Automation Tool (CAT) is being developed as part of an effort to modernize HSI’s Form I-9 Inspection Process. The goal is to uses machine learning (ML) and automation to increase the speed and efficiency of ingesting and processing Forms I-9 data. Easy to use front-end interface workflow that increases work productivity and reduces manual entry. RAVEn CAT currently employs an Optical Recognition Service (OCR) model and software (Tesseract OCR) to identify pixel coordinates of handwritten and read/extract computer typed characters from ingested forms for processing. Additional research into opensource ML Object Detection models is being made to help further augment accuracy of text identification and extraction of ingested forms into the pipeline.
Deployment Status: Inactive (no longer used)
Use Case Name: Data Tagging and Classification
Use Case ID: DHS-416
Use Case Summary: The Homeland Security Investigations (HSI) Innovation Lab is developing an analytical platform called the Repository for Analytics in a Virtualized Environment (RAVEn). RAVEn facilitates large, complex analytical projects to support ICE’s mission to enforce and investigate violations of U.S. criminal, civil, and administrative laws. RAVEn also enables users to develop new tools to analyze trends and isolate criminal patterns as HSI mission needs arise. For more information, please read the DHS/ICE/PIA-055 - Privacy Impact Assessment 055 for the Repository for Analytics in a Virtualized Environment (RAVEn). RAVEn leverages data tagging and classification to do the following: The Email Analytics Tool streamlines how special agents and criminal analysts search, filter, translate, and report on electronic communications evidence and will help investigators more effectively determine the structure and organization of criminal enterprises. The RAVEn - Lead Tracker is a centralized system where agents can send and receive leads and enter outcomes such as arrests and seizures. The goal is for all leads in the agency to be found in one place, rather than in various email inboxes. The overarching goal of Mobile Device Analytics is to improve the efficiency of agents and analysts in identifying pertinent evidence, relationships, and criminal networks from data extracted from mobile devices.
Deployment Status: Inactive (no longer used). This use case had multiple parts that have been separated into individual, inventoried use cases: Email Analytics for Investigative Data (DHS-48) and Mobile Device Analytics for Investigative Data (DHS-49).
No changes to AI use cases within ICE since December 16, 2024. For other updates view the Full DHS AI Use Case Inventory on the DHS AI Use Case Inventory publication library.