Performance Targets for IA EA Detection of Undeclared Nuclear Activities

Under the state-level concept (SLC), the IAEA identifies fo r each state a set of possible acquisition paths by which the st ate could acquire nuclear mate rial for use in a nuclear explosive if, hypothetically, it were to ta ke a decision to do so. To detect such acquisition, the Agency identif ies three generic state-le vel objectives: to detect undeclared nuclear material or activities anyw here in the state; to detect undeclared production or processing of nuclear material in declared nuclear facilities; and to detect diversion of declared nuclear material. When these generic goals are applied at a specific step in an acquisition path, they lead to state-specific technical objectives. As a benchmark of effectiveness, IAEA Safeguard s Department official s have suggested the concept of incorporating a performance ta rget for each technical objective under its guidance for implementing the SLC. For tech nical objectives relating to detection of undeclared activities, however, setting performance targets presents certain challenges. The set of possible undeclared activities an adversary could undertake is large, and the likelihood of detection depends not only on IA EA efforts but also on information from open sources or third parties. A direct causa l link between IAEA activities and detection probability would be difficult to formulate, and this affects the nature of the assurances the Agency can provide. This paper will suggest how performance targ ets for detecting undecl ared activities might be formulated, consistent with the Agency’s important role in detecting, investigating, and deterring such activities. Targets woul d specify what should be done, rather than what detection probability should be ach ieved. One overarching goal would be to maintain a consistent model of the state’ s nuclear and nuclear-related activities and capabilities. Another target, for example, might be to conduct adequate open-source collection and analysis, including broad collection and revi ew as well as focused research and analysis on any priority t opics identified in acquisition pa th analysis. Taking steps to resolve questions and inconsistencies arisi ng from such analysis would be another important performance target.