Azure Adaptive Virtual Patching
introduces an AI-orchestrated Adaptive Shield designed to reduce exposure to
new, emerging, and newly disclosed zero-day vulnerabilities while customers
test and deploy software patches. The initial workload focus is Linux Virtual Machines
(VM) and Virtual Machine Scale Sets (VMSS) on Azure.
The service connects three semantic layers: a knowledge graph of
vulnerabilities and exploit conditions, a map of the customer’s actual
workloads and runtime context, and an orchestration layer that identifies
relevant exposure and generates workload-specific mitigations. AI-synthesized
policies use explicit matching conditions and enforcement actions, combining
adaptive reasoning with deterministic runtime enforcement. The Adaptive Shield
architecture coordinates eBPF-based workload protection and network-based
compensating controls to address relevant attack paths.
Customers onboard workloads as Protected Resources. A workload
extension collects telemetry and deploys mitigations directed by the service’s
AI core, helping protection adapt as threats and workload conditions change.
Azure Adaptive Virtual Patching complements existing security investments and
gives customers time to patch safely. It applies temporary mitigations rather
than installing software fixes; the corresponding mitigations can be rolled
back after patching and confirmed remediation.
Learn more: