Many open collaboration platforms (e.g., Wikipedia) have recently utilized automated algorithmic agents, called Bots, to solve the issue of stagnating user participation. However, understanding of how such a bot agent affects user activities in open collaboration is limited. In this study, we examine (i) whether bot intervention affects user participation and (ii) whether the impact of bot intervention could vary by characteristics of open collaboration works. We pursue our research goals by utilizing a rich dataset between 2005 and 2017 from an online open collaboration platform. The platform has recently utilized two types of unique bots that interact with human users in open collaboration. By employing difference-in-differences approach, we show that bot intervention in open collaboration leads to unintended consequences as it significantly demotivates user participation. We further show that the negative effect of bot intervention is alleviated if more active users are engaged or if covered topics have a higher concentration in a collaboration work. This provides directions for businesses to resolve the unintended demotivation problem in utilizing bot agents.