Opinion formation is one of the most fascinating phenomena observed in human communities, and the ability to predict and to control the dynamics of this process is interesting from the theoretical as well as practical point of view. Although there are many sophisticated models of opinion formation, they often lack the connection with real life data, and there are still sociological processes that need to be explained. To address this, we propose a model describing the dynamics of opinion formation which mimics the process of the virus or disease spreading in the population. The introduced model is motivated by the model of disease spread with three possible channels - direct contact, indirect contact, and contact with "contaminated" elements. We demonstrate that the presence of "contaminated" elements, which in the case of on-line communities can be represented as the content published on the Internet, has considerable impact on the process of opinion formation. We argue that by using a simple mechanism of opinion spreading via passive elements, the introduced model captures the meaningful elements of opinion formation in complex communities. The presented work provides a step towards formulating universal laws governing social as well as physical or technical systems.