This paper discusses various soft security considerations that should be accounted for in the next generation of advanced military unmanned systems. By modeling unmanned system teams as mobile ad hoc networks, we underscore the different types of information-based security vulnerabilities that motivated adversaries may be able to exploit in unmanned systems. Then we provide an overview of computational trust and show that it can be used to defend against these vulnerabilities by finding the most reliable agents to interact with from a pool of potential agents. Finally, we discuss ongoing work at U.S. Army TARDEC that is applying computational trust within a vehicle controller for autonomous convoy operations.