[demetersson]

State-of-the-art artificial intelligence (AI) systems are almost certain to transform how the IC executes its mission, but the risks that accompany the use of such powerful tools are not well understood. Before any AI system is deployed, the IC should focus on understanding AI decisionmaking processes to ensure that they are explainable, align with human ethical standards, and are controllable. As AI systems become more capable and ubiquitous both at home and abroad, the Intelligence Community must fully understand their decision spaces to guarantee that unintended and potentially catastrophic outcomes can be anticipated and interdicted before AI system deployment.