Knowledge Reformulation and Deception as a Defense Against Automated Cyber Adversaries

Leveraging automated planning has been shown to be advantageous for automating network penetration testing, providing a foundation to generate intelligent approaches to attacking a target system. Unfortunately, this same technology has the potential to be abused by actual attackers, presenting a challenge to defenders. In this paper, we investigate how we can leverage ideas from the deception community to reduce the automated planning capacity of an actual attacker. Our extensive experimental analysis sheds some light on the susceptibility of planning-based attackers to knowledge modifications, potentially yielding to new insights on future techniques for cyber defense.