The Department of Homeland Security revealed its first-ever roadmap for AI, outlining key initiatives and policies for current and future AI implementation. The plan includes directives on AI, guidelines on AI security, and strategies to address AI-related risks.
The Department of Homeland Security (DHS) has unveiled its inaugural roadmap for artificial intelligence (AI) on March 18, 2024. The plan, which addresses the agency’s current AI uses and future strategies, includes several key initiatives and policies.
Key aspects of the roadmap include a forthcoming department-wide directive on AI, new guidelines from the Cybersecurity and Infrastructure Security Agency (CISA) focused on AI security, and an anticipated report from the Countering Weapons of Mass Destruction Office regarding AI-related risks. DHS Secretary Alejandro N. Mayorkas emphasized the importance of balancing AI’s potential benefits with its associated risks in a press release.
Applications of AI within DHS involve tracking suspicious vehicle patterns at borders, evaluating building damage post-disasters, and employing large language models (LLMs) for training certain officers. Additionally, U.S. Citizenship and Immigration Services is considering LLMs for training officers and Homeland Security Investigations is exploring AI for document pattern analysis.
The roadmap also includes plans to establish a new working group within the Science and Technology Directorate to develop an action plan addressing algorithm training, pilot projects, and AI-enabled adversaries. There is also an initiative to create a testbed for independent assessment services.
This release follows the agency’s recent moves to strengthen its AI capabilities, including the hiring of 50 AI experts for an AI Corps and the formation of an AI task force last year. DHS has periodically updated its public AI use case inventory, as mandated by a Trump-era executive order.
