This paper presents a model of interaction between anticorruption and corruption bodies. The main policies of anticorruption authorities are aimed at reducing the level of corruption in some entities through optimal allocation of resources and effective anticorruption measures. A model based on Markov decision-making process was developed. Howard's policy-improvement algorithm was used to find an optimal decision strategy. The numerical example is presented in which the real parameters for corruption level estimation called "Red Flags" were used.