Introducing Enhanced AuthZed Tools for Permission Management in RAG and Agentic AI Systems

AuthZed Inc., a permissions management startup, has unveiled support for Retrieval-Augmented Generation (RAG) and Agentic AI systems, enhancing its infrastructure to tackle security issues in enterprise AI. This initiative provides engineering teams with essential tools to ensure AI systems adhere to permissions, avert data leaks, and maintain comprehensive audit trails. With organizations facing increasing complexities in authorization due to interactions between AI systems and proprietary data, traditional authorization methods are inadequate. AuthZed’s open-source system, SpiceDB, built for scalability, addresses these challenges by efficiently processing vast access control lists and authorization checks. The system enables teams to enforce secure operations through document filtering and real-time permission synchronization with platforms like Google Workspace. Moreover, the Agentic AI Authorization Model governs agent actions, aligning them with user permissions while including controls for functionality, access limitation, and oversight. AuthZed has raised $15.9 million in funding, reflecting the growing need for secure AI project infrastructure.

Source link