A system of ordinary differential equations (ODEs) was developed to model the interaction dynamics between public servants and the citizens they serve to provide insights to the evolution of corruption in the public service. Corruption is modeled as a contagious disease that can "infect" susceptible citizens when they interact with "infected" (i.e., corrupt) public servants through harassment bribery. In this model, the public servants and the citizens are compartmentalized into different classes. The public servants could be honest (SH), on a crossroad (SX), corrupt (SC), or dismissed (SD), while citizens could either be upright (CU), apathetic (CA), cooperators (CC), or whistleblowers (CW). SH do not ask for harassment bribery while SX have asked for harassment bribery but only choose which citizen to ask from. SC always ask for harassment bribery no matter what, while SD have been dismissed from public service (and prosecuted) as a result of complaints. CU and CA do not give bribery but the former always complain when public servants ask for them, while the later do not do anything. CC and CW give bribery but the former willingly give them, while the later grudgingly give them and then complain. The ODEs provide for intra- and inter-action dynamics between the public servants and citizens with probabilities of interactions among actors as parameters of the model. Various combinations of interaction probability values provide for scenarios when corruption in the form of harassment bribery will become contagious, epidemic, and self-correcting (i.e., self-healing).