Chapter 3 of 6 · Before you start

Azure Policy guardrails for Microsoft Foundry AI services

Optional module · Draft 4 hours

Chapter 3 of 6

Confirm these prerequisites:

  • The approved target scope is the subscription that contains the Foundry and Azure AI services resources governed by this module.
  • The deployment operator has time-bound Resource Policy Contributor on the subscription where the initiative and assignment are deployed.
  • The operator can run subscription-scope Bicep what-if.
  • A Log Analytics workspace exists, and its owner can approve the assignment identity's access before diagnostic remediation moves to DeployIfNotExists.
  • The platform policy owner, compliance review owner, data-residency owner, and promotion authority can review current policy states together.
  • The team knows which controls are owned by the related sessions named in the boundaries.

Implementation files#

TypeFileConsumer
Deploymentartifacts/policy/definitions/restrict-foundry-deployment-sku.jsonThe subscription policy deployment pipeline
Recordartifacts/policy/guardrail-decisions.jsonThe policy deployment pipeline, preflight scripts, and compliance review owner
Deploymentartifacts/policy/initiative.bicepThe subscription policy deployment pipeline
Deploymentartifacts/policy/assignment.bicepThe subscription policy deployment pipeline
Deploymentartifacts/environments/initiative.bicepparamThe subscription policy deployment pipeline
Deploymentartifacts/environments/assignment.bicepparamThe subscription policy deployment pipeline

Resolve every __REQUIRED_*__ value before running a live preview. Keep tenant IDs, subscription IDs, policy-state exports, resource IDs beyond the approved scope, and customer data in customer systems.

Optional module

Azure Policy guardrails for Microsoft Foundry AI services slide deck