Securing Federal LLM Systems: Multi-Agent Workflows and Data Safety

This conference talk from RSA Conference explores secure architectures for deploying large language models in federal environments, comparing cloud-based and on-premise solutions that meet FISMA standards. Learn about the risks associated with fine-tuning models, the advantages of retrieval-augmented generation (RAG), and how multi-agent workflows can enhance security and compliance. Presented by Robert Buccigrossi, CTO of TCG, the 48-minute session provides real-world examples from federal projects to demonstrate practical implementation of secure LLM systems in government settings.