ConfidentialRAG by SafeLiShare is a confidential analytics platform introduced as a private beta solution at RSA Conference 2024. It is designed to enable secure and privacy-preserving Ground Retrieval Augmented Generation (RAG) workflows in Large Language Models (LLM) using an organization's data, vector databases, or knowledge graphs. The platform aims to protect chat prompts, responses, and history from unauthorized access. ConfidentialRAG utilizes secure enclaves, hardware-based encryption, and confidential computing technologies to process applications and data, preventing exposure of unencrypted information at runtime. It ensures policy-driven access to applications and data within trusted execution environments, guaranteeing data privacy and fulfilling data governance and compliance regulations.
