Administrative corruption poses a significant challenge within public and organizational operations. This paper proposes a simple decision support system to estimate the probability of individual corruption and the level of societal corruption, and to use these estimates for monitoring and detection over time. Each period (e.g., monthly), the organization updates the corruption probability of every person using two inputs: (a) relatively stable personal features from the World Values Survey (WVS), and (b) live organizational signals from the internal whistleblowing system, third-party sources, observed suspicious actions, and proximity to confirmed cases. To reflect the wider environment, we also include a public context signal: a society-level corruption-risk forecast derived from countries’ Sustainable Development Goals (SDGs) indicators. We analyze three types of data from international databases using standard data analysis and machine-learning methods and report model performance. To examine effectiveness, we simulate a small, hypothetical organization using profiles derived from the WVS dataset and evaluate the approach under controlled conditions. The proposed approach flags anomalous individuals and detects system-level abnormalities within a reasonable timeframe, while keeping the process transparent, periodic, and feasible under realistic inspection constraints.