The U.S. Nuclear Regulatory Commission (NRC) is advocating for ongoing system monitoring and robust safeguards to ensure strong boundaries for AI tools.

In a recently released document, the NRC, in collaboration with the United Kingdom’s Office for Nuclear Regulation and the Canadian Nuclear Safety Commission, outlined potential requirements for the use and governance of AI in the nuclear sector.

“AI could support outcomes that are not readily attainable via non-AI techniques. For example, AI could analyze a much larger volume of data than conventional approaches, potentially allowing better management of plant risks and improved efficiency. Similarly, sufficiently developed AI could accomplish tasks that have until recently only been possible using humans, reducing the need to enter hazardous areas, and potentially reducing error,” the document said.

Moreover, AI’s capacity to be retrained with updated information allows it to quickly learn from past experiences and improve over time. 

Its distinctive capabilities enable tasks that conventional technologies or humans cannot achieve, opening up opportunities to enhance safety, security, and efficiency both directly and indirectly.

The commission believes AI safety and consistency can be ensured by setting boundaries, including data restrictions, controlled software interactions, and using diverse, redundant, and isolated systems to minimize unintended actions.