REGMAR (MarineTime Regulation): LLM-Based RAG & Multi-Agent AI for Maritime Cybersecurity

The maritime industry is undergoing rapid digital transformation, driven by Maritime Autonomous Surface Ships (MASS) as regulated by the International Maritime Organisa- tion (IMO). This transformation has introduced unprecedented cybersecurity challenges. To address these issues, the IMO and the International Association of Classification Societies (IACS) have introduced various cybersecurity regulations, which are interpreted and implemented through diverse approaches by individual classification societies. However, the maritime sector continues to face significant challenges in the application and interpretation of these regulations. This study proposes a multi-agent and Retrieval-Augmented Generation (RAG) based framework, designed for integration with generative AI models, to enable the effective application of cybersecurity regulations issued by various classification societies using diverse large language models (LLMs). The practicality and applicability of the proposed framework have been demonstrated, highlighting its potential to mitigate the difficulties faced by the maritime industry in complying with cybersecurity regulations. Furthermore, this research contributes to the accelerated adoption of robust cybersecurity measures across the maritime sector.