© 2018, Springer Nature Switzerland AG. We examine the security of collaboration systems, where several users access and contribute to some shared resource, document, or database. To protect such systems against malicious servers, we can build upon existing secure messaging protocols that provide end-to-end security. However, if we want to ensure the consistency of the shared data in the presence of malicious users, we require features that are not available in existing messaging protocols. We investigate the protocol failures that may arise when a new collaborator is added to a group, and discuss approaches for enforcing the integrity of the shared data.