Gravity of Artificial Intelligence (AI) Large Language Models (LLM) on Offensive and Defensive Approaches
Karthik R Sundar
Director of Cyber Risk Advisory, Deloitte Malaysia
Session Outline
This session provides a background on how AI LLM aids offensive cyber techniques and tactics, and, on the other hand, may provide equal opportunity for defenders to leverage the technology to adapt and double down on proactive defensive measures.
The session will discuss:
-
Cyber Prompt Engineering - a new type of tactic, technique, and procedure (TTP) that can be leveraged for offense
-
Social engineering techniques enabled by AI – how AI can be used by the offense to further exacerbate the age-old problem of ‘the attack on the human’
-
Bypassing guardrails on LLM – how we can bypass guardrails implemented in LLM, that aim to avoid inaccuracies or provocative activity
-
Defensive model patterns enabled by LLM – using LLM for cyber defense