AI Tools Speed Up Compliance and Research Tasks in US Federal Agencies

Federal agencies are using artificial intelligence to accelerate compliance analysis and research tasks, while frameworks and oversight focus on transparency and explainability in these processes.

Cybersecurity officials from the Department of the Air Force and the State Department’s Bureau of Diplomatic Security stated that AI systems are being used to summarize compliance outputs, allowing human analysts to review information more efficiently. These tools support analysts by converting complex data into bullet-point summaries, which can streamline decision-making and task completion.

Lt. Col. Frank Jamerson described a process where compliance analysts use AI-generated outputs to indicate compliance and then organize these findings into concise summaries. Manny Medrano of the State Department’s Bureau of Diplomatic Security said that AI enables analysts to handle larger volumes of data, which allows them to dedicate more attention to tasks that require greater complexity and judgment.

NORC at the University of Chicago has developed a framework, known as AI-DQSI, that guides the responsible use of AI in federal statistical data preparation. This framework emphasizes the importance of transparency, explainability, and human-in-the-loop design to ensure that users can understand and intervene in AI-generated outputs when necessary.

What the numbers show

• Federal agencies reported about 1,200 current and planned AI use cases
• Only five out of 23 civilian CFO Act agencies provided complete information for each AI use case
• GAO identified incomplete and inaccurate inventories as a management challenge

The NORC framework also highlights how AI can assist with tasks such as metadata creation, data harmonization, and quality assessment, while maintaining analyst oversight and control. This approach is intended to balance the efficiency benefits of AI with the need for human supervision and accountability in federal data processes.

Despite the growing adoption of AI tools, the Government Accountability Office (GAO) reported that inventories of AI use cases across federal agencies are often incomplete or inaccurate. According to the GAO, only a small number of agencies have provided comprehensive details for each AI application, which can hinder effective management and oversight of AI deployment.

Federal agencies have identified approximately 1,200 AI use cases, both current and planned, across 20 of the 23 civilian CFO Act agencies. However, the GAO found that incomplete reporting and data gaps remain a persistent issue, affecting the ability to monitor and guide AI implementation effectively.

The GAO stated that these shortcomings in AI use case inventories present challenges for managing the deployment and oversight of AI technologies in federal operations. Efforts to improve the accuracy and completeness of these inventories are ongoing, with the aim of supporting responsible and effective AI use across government agencies.

* This article is based on publicly available information at the time of writing.