The Department of Homeland Security (DHS) has released an AI roadmap that details its plans to implement artificial intelligence (AI) across its operations. The roadmap highlights the department's intention to use generative AI and large language models (LLMs) for training immigration officers, aiding law enforcement investigations, and improving disaster mitigation plans. The DHS will establish an "AI sandbox" for responsible testing and expand it to more users over time. The department will also create a test and evaluation infrastructure for AI, establish an AI Safety and Security Board, issue a new enterprise-wide AI policy, and recruit 50 AI experts by the end of 2024.