Enhancing Homeland Security Operations with GenAI
Every month, artificial intelligence (AI) increases in both popularity and capability. AI refers to a computer’s ability to perform automated tasks in a manner that mimics human intelligence. In today’s fast-paced environment, AI provides Federal agencies with a solution to reduce time spent on repetitive tasks (e.g., data entry and analysis) to prioritize complex, mission-critical work.
The Department of Homeland Security (DHS) was one of the first Federal agencies to adopt this technology. In 2024, DHS unveiled a comprehensive plan to integrate Generative AI (GenAI) into the agency’s operational components. GenAI is a subset of AI that is trained on existing patterns to create new content including summaries, videos, images, and audio.
DHS launched three GenAI pilots to evaluate the benefits and limitations of this technology in an immigration, emergency preparedness, and law enforcement capacity.
Immigration: U.S. Citizenship and Immigration Services (USCIS) identified the need to improve the agency’s training capacity. They created a pilot program to enhance USCIS officer training with the aid of large language models (LLMs). In this pilot, officers practiced eliciting testimony from AI-generated refugees and asylum seekers. Officers commended the training tool’s accessibility and how the tool gave them the flexibility to complete the training at their own pace and schedule
Emergency Preparedness: The Federal Emergency Management Agency (FEMA) recognizes that developing community hazard mitigation plans is a challenging and time-consuming process for state and local Government agencies. For their GenAI pilot program, FEMA used a LLM to generate drafts of hazard mitigation plans that were customizable to each participating agency’s needs
Law Enforcement: Homeland Security Investigations (HSI) wanted to optimize the efficiency of their investigative summaries and processes. HSI’s GenAI pilot program also used LLMs and an open-source AI model to generate summaries of law enforcement reports. The program enabled law enforcement officers to analyze millions of reports for relevant information with the aid of a semantic search engine. The pilot successfully demonstrated the benefits of using GenAI to enhance the efficiency and accuracy of investigations
These three pilot programs highlight the scope of GenAI in Government, from training Federal employees to distilling complex datasets into accessible, high-level summaries. However, emerging technologies like GenAI come with certain risks. Without proper risk mitigation measures in place, nefarious actors can abuse AI to conduct attacks on our nation’s critical infrastructure and spread misinformation campaigns that undermine trust in U.S. elections. Therefore, it is critical that Federal agencies implement effective risk mitigation strategies when integrating AI into their processes.
