Implementing authorization in RAG-based AI systems with Cerbos

Cerbos introduces a new access control feature for Retrieval Augmented Generation (RAG) and Large Language Models (LLMs), helping software builders secure their AI applications. This solution addresses the challenge of providing AI models with context while ensuring privacy and authorization policies are not violated. Cerbos' functionality applies row-level data filtering before data is passed to the LLM, ensuring responses are tailored to user access privileges. It reconciles the need for vast data access in AI with the Zero Trust principle of least privilege access.