The Defense Nuclear Facilities Safety Board (DNFSB) is setting terms on how it can start using generative artificial intelligence (AI) to enhance nuclear safeguards “without undue risks.”
The policy is laid out in a Dec. 29 directive that will apply “to all DNFSB personnel, it full-time employees, contractors, consultants, interns, and any individuals granted access to DNFSB systems, data, or networks.”
Importantly, the policy states DNFSB officials are not merely sitting back and watching the machines take over, but rather will verify the information that results from generative AI.
Generative AI is generally considered artificial intelligence that upon a user’s request can make original content that might include images, text, video or software code. Industry and government are increasingly interested in how to harness such technology for beneficial uses.
“To deploy AI tools in a manner that preserves public trust, the DNFSB will always ensure agency decisions and final work products are prepared and reviewed by qualified experts and responsible managers,” according to the directive.
The board’s executive director of operations, Mary Buhler, will appoint a chief AI officer to coordinate all DNFSB activities, according to the directive.
DNFSB users are responsible for verifying the accuracy of information gleaned from generative AI systems.
In 2025, the Trump administration published guidance for federal agencies on the adoption of AI, a DNFSB spokesperson said via email. Key documents on this include Executive Order 14179, Removing Barriers to American Leadership in Artificial Intelligence, as well as a memo on AI use by the Office of Management and Budget.
The Defense Nuclear Facilities Safety Board was created by Congress to provide independent safety analysis and advice to the Department of Energy on DOE nuclear installations.