Augmented reality refers to overlaying computer-generated data on real world senses, as well as possibly taking continuous input from camera, microphone, or other sensors. Several such systems are shipping today, notably the Kinect and iPhone applications such as Layar, Wikitude, and Junaio. The Google Glass heads-up display project has announced that it will ship the first developer kits in early 2013. The area has also seen sustained academic attention. Despite this activity, relatively little attention has been paid to the security issues with multiple applications sharing one augmentation platform, a core OS concern. We identify security issues that are specific to augmented reality systems. We then propose an OS architecture for multi-application augmented reality that enables a clean separation of concerns. Finally, we give two example approaches that each address a subset of these issues and our experiences building prototypes of both. The time is right for the systems community to tackle the question of how multiple applications should share ``one (augmented) reality."