Communication facilitates cooperation by allowing deviators to be widely punished. This paper explores how players might misuse these communication systems to threaten one another and identifies conditions under which communication improves cooperation despite these threats. A principal plays trust games with a sequence of short-lived agents who communicate. An agent who shirks can still demand pay by threatening to report that the principal deviated. We show how these threats can destroy cooperation. Cooperation is partially restored if there are public signals of the agents' efforts or the principal's transfers, or if the principal has ongoing bilateral relationships with each agent.