Microsoft AzureOverview
Vector stores and AI search indexes frequently contain embeddings, document chunks, retrieval metadata, or business knowledge used by AI systems. Public access can expose sensitive retrieval data and increase prompt-injection and data-extraction impact.
Remediation guidance
Remediation
Disable public access to AI search and vector-store services where possible. Use private networking, scoped identities, query-key rotation, and request logging.
- Move access to private endpoints or approved gateways.
- Review index keys and callers.
- Remove anonymous or broadly shared query access.
Query logic
These are the stored checks tied to this control.
AI vector and search stores should not be public
Connectors
Microsoft Azure
Covered asset types
AI ServicesSearchVector Store
Expected check: eq []
{
azureAISearchServices(where: { publicEndpoint: { eq: true } }) { ...AssetFragment }
}
