A hardware-level code vulnerability detection system powered by self-tuned Large Language Models.
In response to the growing complexity and vulnerability of hardware-level code, particularly in encryption methods, this study introduces an innovative detection system powered by self-tuned Large Language Models (LLMs). The primary objective was to develop a system capable of accurately identifying vulnerabilities in hardware-level code snippets, with a specific emphasis on encryption techniques like AES. The methodology entailed collecting a diverse dataset of both vulnerable and non-vulnerable code snippets, which served as the training material for a general LLM. This LLM was then fine-tuned using the collected dataset, enabling it to understand and analyze the intricacies and potential security flaws inherent in hardware-level code.