Toward Leveraging Large Language Models to Support Cybersecurity Risk Assessments in Industrial Component Development

As the global cybersecurity threat landscape continues to evolve, the importance of implementing robust cybersecurity risk management increases for organizations of all sectors and industries, including component manufacturing. Cybersecurity risk assessments form a fundamental basis for the mitigation of such risks and need to be integrated early and throughout the component development life cycle. However, these assessments are currently performed manually and require substantial financial and human resources. Assessment results may also vary in quality due to a reliance on the individual expertise of human security experts. To increase efficiency and consistency, this work presents an approach to support cybersecurity risk assessments during the development of industrial components by leveraging large language models and retrieval augmented generation. Two main application cases are identified for the integration of such models within established industry workflows. Additionally, a prototypical software implementation is presented that can be integrated seamlessly into existing risk assessment activities.