Recently, the Department of Homeland Security, or DHS, announced its plan to integrate and deploy Artificial Intelligence across various DHS missions. The agency released a roadmap, detailing an aggressive set of goals, with multiple targets for the end of 2024.
The DHS is the first major U.S. government agency to substantially implement AI in its policies and plans, spurred by the "unprecedented speed and potential of AI's development."
An overview of DHS's goals
The DHS announcement targeted national security, improved operations, and efficiency as overarching goals, and anticipate expanding the use of AI substantially across all aspects of US homeland security apparatus.
The Roadmap also stressed the need for responsible use of AI, and stated that it will ensure its use of AI (1) is responsible and trustworthy; (2) is rigorously tested to be effective; (3) safeguards privacy, civil rights, and civil liberties; (4) avoids inappropriate biases; and (5) is transparent and explainable to its workforce and to those it serves.
The DHS AI Roadmap lays out three major "Lines of Effort" and identifies numerous "Workstreams" within each Line of Effort. The Lines of Effort are described as follows:
- Responsibly Leverage AI to Advance Homeland Security Missions While Protecting Individuals' Privacy, Civil Rights, and Civil Liberties
- Promote Nationwide AI Safety and Security
- Continue to Lead in AI through Strong, Cohesive Partnerships
Current and future uses of AI
The Roadmap notes that DHS already uses AI in several ways, from drug traffic detection to TSA processes. In the future, DHS plans to deploy AI in other ways, including efforts to root out cyberattacks targeting critical US infrastructure.
The Roadmap also described three pilot programs: combatting child sexual exploitation (Homeland Security Investigations), developing hazard migration plans (Federal Emergency Management Agency), and improving officer training (United States Citizenship and Immigration Services). These diverse programs suggest the ambitious scope of DHS plans for AI, and were developed through the evaluation of Large Language Models, or LLMS, and Generative AI programs available to the Department.
The DHS also announced it intends to lead in AI through partnerships with organizations ranging from private businesses to state government agencies to non-profits, with the expectation that insight from the DHS AI programs will help public and private entities adopt AI in their operations.
Addressing AI's risks and challenges
The DHS Roadmap also acknowledged several concerns with the development and use of AI. It stated one of AI's biggest threats is that it will, undoubtedly, be utilized by nation-state adversaries and cyber criminals, which presents a particular threat to critical infrastructure and the organizations that maintain it. Additionally, AI programs cannot distinguish inherent problems within data sets, and the results of certain AI processes could lead to inappropriate or biased outcomes.
DHS is testing an 'AI sandbox' to allow for careful analysis of and experimentation with LLMs in safe conditions. Since AI effectiveness is heavily based upon the size and quality of its data sets, not all of which are acquired with consent from subjects, AI's privacy concerns have been well documented internationally. AI has supercharged many industries, including cybercrime, and even well-intentioned usage poses risks.
What does this mean for me?
It may mean very little. While you may have been exposed to AI-augmented processes in airports, for example, it's not clear how fast or how public the other DHS use of AI will be. Even expansive use of AI may remain mostly a government concern for the time being. However, what is clear is that AI is of major interest to the U.S. government, from the White House's broad new goals to a slew of legislative proposals, and likely many more new applications in coming years.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
