The U.S. Nuclear Regulatory Commission (NRC) is seeking information from vendors about the use of artificial intelligence (AI) to detect cyberattacks on power plants.
In a sources sought notice posted to SAM.gov, NRC said it is looking for qualified vendors, including educational institutions, to provide all management, supervision, administration, and labor for the requirement entitled, “Characterizing Nuclear Cyber Security Using Artificial Intelligence/Machine Learning.”
NRC is planning to conduct research to prepare for the regulation of nuclear applications of AI and ML, and wants a vendor capable of doing that research. Among the tasks to be performed as part of the research include:
- Identifying technologies and approaches applicable to AI and ML use cases, especially those that capture and analyze system and cybersecurity states of nuclear plant systems;
- Evaluating available technologies and approaches, and selecting a technology or approach for research on and demonstration of the AI and ML use case;
- Implementing the “selected approach with a small test case to develop insights and fundamental knowledge about the application of AI/ML to nuclear cybersecurity such as the feasibility of detecting and distinguishing abnormal plant and cybersecurity states, considerations for accuracy and reliability, necessary and availability ML training data, utility for identifying and responding to a cyberattack and existing and novel risks inherent to the application of the technology;” and
- Producing a technical report documenting the project results and generalizing the results of the test case as appropriate.
This isn’t NRC’s first foray into researching how AI and ML for different applications. Last year, the agency released a request for comment on how it could utilize AI and ML in nuclear power generation. The agency sought comment from commercial nuclear power companies for that inquiry, published April 21.