Asymmetric LLM-aided Cyber Effects

Recent advances in AI/ML technologies have presented an opportunity to address challenges in reliably and efficiently performing Defensive and Offensive Cyber Operations (DCO and OCO, respectively).Ý Traditionally, the use of dedicated defensive (ìblueî) and adversarial (ìredî) teams was employed resulting in significant human-in-the-loop costs and latency while also requiring best-in-breed expertise to analyze todayís complex DoD systems ñ often resulting in an exhaustive evaluation being intractable. Ý The proposed solution aims to address these hurdles by fine- tuning state-of-the-art Large Language Models (LLMs) to support both DCO and OCO in a unified solution.