As autonomy becomes ubiquitous, it needs a way to reason strategically about other agents in its environment. When these agents operate with incomplete information about the environment and other agents, an autonomous agent could use deception to gain an advantage. In this talk, we propose a hypergame theoretic approach to automatically synthesize deceptive strategies for adversarial two-player interactions over graphs under various classes of incomplete information. Specifically, I will discuss how to characterize the states from which the autonomous agent can gain an advantage and use this information to synthesize a deceptive strategy. Moreover, I will discuss how our hypergame theoretic approach can be combined with formal methods to solve the decoy allocation problem, which involves finding the optimal placement of decoys to protect a cybernetwork. This talk will present our main contributions and discuss future directions for research in the field of autonomous agent deception.