The research program focuses on the systematic analysis of the potential of Large Language Model (LLM)-based tools in software analysis. Its goal is to investigate their use in tasks such as vulnerability detection, static and dynamic analysis, reverse engineering, and code obfuscation/de-obfuscation. The project will critically review both academic research and industrial solutions, evaluating their effectiveness, limitations, and application contexts. Particular attention will be given to the results achieved in existing studies, highlighting successes, challenges, and unresolved issues. Another important research area concerns prompt engineering and how different prompt formulations affect the quality, reliability, and reproducibility of LLM-generated outputs. Finally, the program aims to identify future development directions for the use of LLM technologies in software analysis.
