Procedural story generation has become an important research area due to its attractiveness to computer entertainment applications. Several approaches have been proposed to create procedural stories for games based mainly on logic programming and decision trees. These approaches use a set of possible events that are executed based on how the player interacts with the story. However, these approaches commonly use a fixed set of characters with predefined relationships. In contrast, in this article, we propose an approach that creates dynamic stories based on two networks: a handcrafted event network and a dynamic artificial social network created for each new story. This allows each story to have its own set of characters with a unique relationship network, resulting in an even wider range of possible stories. We demonstrate this with some examples and discuss how this approach can be implemented for interactive computer entertainment systems, such as video games.