"LLM meets Malware: Starting the Era of Autonomous Threat"

Researchers at B42 Labs have shared some findings from their exploratory research on the application of Large Language Models (LLMs) to malware automation, examining how a potential new type of autonomous threat may manifest in the near future. The researchers explored the potential architecture of an autonomous malware threat based on four main steps: Artificial Intelligence (AI)-assisted reconnaissance, reasoning and planning, and AI-assisted execution. They demonstrated the possibility of using an LLM to recognize infected environments and determine which malicious actions would be most appropriate for the environment. In order to leverage LLMs in the complex task of generating code on the fly to accomplish the malicious objectives of the malware agent, they adopted an iterative code generation strategy. This article continues to discuss findings from B42 Labs researchers' analysis of the application of LLMs to malware automation.

Security Affairs reports "LLM meets Malware: Starting the Era of Autonomous Threat"