Audit: DOE-OIG-25-07

December 9, 2024

Progress Is Needed to Support the Department of Energy’s Integration of Artificial Intelligence Into Intelligence Activities

The Intelligence Authorization Act for Fiscal Year 2023 (Act) contains several requirements for the Director of National Intelligence (DNI) and Federal intelligence agencies to implement artificial intelligence (AI) capabilities into intelligence operations.  This includes creating policies, standards, guidance, and implementation plans, and reporting to Congress.  The Act also requires associated Offices of Inspector General to conduct audits of the implementation of those requirements within 2 years of the Act’s enactment.

We conducted this audit to determine the actions taken and/or barriers impacting the Department of Energy’s integration of AI into its intelligence activities.

The Act established requirements for Federal intelligence organizations, including the Department’s Office of Intelligence and Counterintelligence (Intelligence), to implement AI capabilities.  However, of 13 requirements identified in the Act, only 1 had been implemented by Intelligence.  The Department cannot implement 11 of the remaining 12 requirements until the DNI takes action to create policies, other guidance, and authorities for intelligence organizations.  To its credit, Intelligence had taken positive steps toward implementing AI, such as planning for several AI technologies and appointing a dedicated official to oversee all AI efforts.  Without required actions completed by the DNI, Intelligence is limited in the actions that can be taken.

Although Intelligence is awaiting guidance from the DNI before it can complete most of the Act’s requirements, there are still many risks associated with AI technologies to which Intelligence could be exposed.  For instance, Intelligence advised us about internal difficulties within its control, such as addressing funding resources and human capital needs, focusing on aging information technology infrastructure, and creating policies and plans for implementing AI.  In addition, it is important that Intelligence plan for and implement AI technologies with defined and clear “guard rails” in accordance with recently issued Federal guidance and ethical standards.

To address the issues identified in this report, we made three recommendations that, if fully implemented, should improve Intelligence’s efforts to integrate AI into its operations.  We made recommendations related to reporting to Congress on efforts to implement AI, focusing on funding needs to fully implement AI, and ensuring that all officials are trained and knowledgeable about AI concepts and application.