The text discusses the challenges of securing data in Retrieval Augmented Generation (RAG) systems, particularly with regards to access control and data duplication. Enterprises struggle to implement granular user-specific access controls due to the diversity of access control modes and the complexity of managing them effectively. To address this issue, Caber Systems proposed a solution that secures data at the chunk level by constructing an index on the side to map relationships between chunks and their sources. This approach provides visibility into each chunk's origin and associated permissions, enabling deterministic assignment of permissions. The system is integrated with LLM workflows using an SDK, allowing seamless access control management, and provides accountability and audit capabilities through detailed observability of application flow.