IoT Software Vulnerability Detection Techniques through Large Language Model

The explosion of IoT usage provides efficiency and convenience in various fields including daily life, business and information technology. However, there are potential risks in large-scale IoT systems and vulnerability detection plays a significant role in the application of IoT. Besides, traditional approaches like routine security audits are expensive. Thus, substitution methods with lower costs are needed to achieve IoT system vulnerability detection. LLMs, as new tools, show exceptional natural language processing capabilities, meanwhile, static code analysis offers low-cost software analysis avenues. The paper aims at the combination of LLMs and static code analysis, implemented by prompt engineering, which not only expands the application of LLMs but also provides a probability of accomplishing cost-effective IoT vulnerability software detection.