This work is carried on within the frame of the EU Horizon project “BAG-INTEL”. The BAG-INTEL project final target is to develop a Risk-based AI tool to support customs intrusive inspection decision making relative to luggage control in airports. The risk assessment component is the only component of the BAG-INTEL Tool that is presented in the paper. The risk assessment component is one of many other components to be developped in BAG-INTEL Tool. It is powered by an algorithm mixes probabilistic statistics (deductive approach) and Analytic Hierarchic Process (AHP) classification theory (inductive approach). Subsequently, the risk assessment algorithm does not exploit any AI methodology or facility. The exploited data are exclusively numerical figures. Hence, the risk assessment algorithm does neither retrieve nor process any personal or biometric data by its concept. Two risks are identified and assessed. The first is to miss a bag containing illegal substances (detection failure). The second is to intrusively inspect a clean bag (false detection). We consider three individual risk indicators to estimate the likelihood of illegal substances in a bag: external-intelligence informative risk indictor, sniffer dog risk indicator, and X-scan risk indicator. The list of searched illegal substances of interest is purposely limited to narcotics, tabaco, currency notes, gold, and polymers. The individual risk indicators should be aggregated to determine one global risk indicators by the risk assessment algorthim presented in the paper. The mathematical models used are briefly presented and applied using academic data simulating the three risk indicators cited above. The paper presents the state of progress in the development of the risk assessment algorithm and the results of its preliminary numerical testing. The paper is intended to be didactic and accessible to Customs Inspection Decision Making professionals as far as the the probabilistic risk assessment concept and the related algorithm are concerned.