Abstract: Malicious Software (Malware) is a major threat to computer systems. While the malware infection problem has been promisingly tackled via Machine Learning (ML) approaches, this approach's viability has been challenged by the construction of Adversarial Examples (AEs) that can bypass ML-based detectors, including AntiViruses (AVs). The State-Of-The-Art (SOTA) in AE generation is the use of Generative Adversarial Networks (GANs), but it presents a major drawback: adding or removing functions from compiled binaries is far from trivial. However, this game is about to change. This talk presents how LLMs can be used to generate functional code to evade AVs while guided by GAN's outputs. Botacin's recent research shows how GANs can be used as prompt generators for LLMs like ChatGPT and how to weaponize GitHub CoPilot to become a malware writing assistant. The talk details the implementation of 3 GANs and their use to generate thousands of malware samples that evaded real AVs.

Bio: Marcus Botacin is an assistant professor in the computer science and engineering department at Texas A&M University. He holds a Ph.D. in Computer Science (Federal University of Paraná, Brazil, 2021), a master’s in computer science (University of Campinas, Brazil, 2017) and a bachelor’s in computer engineering (University of Campinas, Brazil, 2015). Botacin’s main research interests are malware analysis and reverse engineering. Botacin’s research has been published in major scientific venues (e.g., ACM Transactions and USENIX Security). Botacin has spoken at academic, industry, and hacking conferences (e.g., USENIX Enigma and HackInTheBox).