Spontaneous communications for either personal or shared activities are becoming very common in wireless pervasive computing environments. However, there still remain challenging problems in this type of communications, especially their security over ad-hoc wireless channels. In this paper we present an approach to using a visual code system in order to establish secure spontaneous communications among wireless devices in pervasive computing environments. After briefly discussing the visual code system, we describe two distinct models of spontaneous communications, and then discuss how the visual code system can be leveraged in order to secure spontaneous communications in those models. We also demonstrate our approach through a proof-of-concept implementation.