Developing Your Own Local LLM (GenAI) for Cybersecurity GRC (L13b)
Public AI tools like OpenAI’s models pose security risks when handling sensitive cybersecurity Governance, Risk, and Compliance (GRC) data, as sending information to external servers may violate privacy and compliance policies. Building your own local large language model (LLM) offers a safer alternative but is often complicated by technical issues like compilation errors and difficulties in training or customizing models with local data.
In this session, I will share a practical, hands-on approach using Retrieval-Augmented Generation (RAG) to overcome these challenges. Instead of retraining, RAG allows you to dynamically inject local context and proprietary information into a pretrained local LLM, enabling up-to-date, relevant AI responses without heavy technical overhead.
This workshop-style guide introduces a straightforward tool and workflow I found useful for deploying local LLMs tailored to cybersecurity GRC. Attendees will learn how to bypass common technical barriers and easily add their own data to improve the model’s relevance for compliance and risk tasks.