Policy-aware Vector Search: A Vision for Fine Grained Access Control in Vector Databases

The article presents a vision for Policy-aware Vector Search, addressing the limitations of fine-grained access control (FGAC) in vector databases, particularly in security-sensitive applications. It formalizes the FGAC policy model for vector databases and discusses the challenges of balancing policy enforcement with approximate nearest neighbor (ANN) search recall and query latency. This work is significant for practitioners as it outlines potential strategies for implementing FGAC in vector databases, which is crucial for secure AI applications.