What happened
Microsoft published CVE-2026-56167 (CVSS 8.5) on 2026-07-23/24, disclosing an SSRF vulnerability in Azure AI Search that permits an authorized attacker to escalate privileges over a network.
Why it matters
Azure AI Search underpins retrieval-augmented-generation (RAG) pipelines for many enterprise Copilot and custom LLM deployments; an SSRF-driven privilege escalation in this shared multi-tenant service could allow lateral movement into backend data stores or internal Azure services that host proprietary embeddings, indices, and enterprise data feeding LLM applications.
Attack vector
A server-side request forgery (SSRF) flaw in Azure AI Search allows an already-authorized attacker to make the service issue requests to internal/unintended network endpoints, which can be leveraged to elevate privileges over the network — a classic cloud-metadata/internal-service pivot pattern in RAG-serving infrastructure.
Affected systems
Microsoft Azure AI Search (cloud service)
Mitigation
Microsoft has issued a fix as a cloud service update (no customer action required beyond staying current); review Azure AI Search access controls and network configuration per MSRC advisory.