Control: Azure > AI Foundry > RAI Policy > Content Filters

Check or enforce the content filters configured on Azure > AI Foundry > RAI Policy resources.

The control compares the RAI policy's content filter list and mode against the sub-policies under Azure > AI Foundry > RAI Policy > Content Filters: one per filter on the input (Prompt) and output (Completion) side, a severity threshold for each harm category, and the policy mode. Only filters whose sub-policy is not Skip are checked, and each asserts only the fields it names. Missing or misconfigured filters put the control in alarm, and its details list every filter with the current and expected value. On Enforce: Configured the control runs the Set Content Filters action, which adds or corrects only those filters and preserves every other filter and setting.

System-managed RAI policies (Microsoft.Default, Microsoft.DefaultV2) cannot be modified in Azure and are skipped.

Resource Types

This control targets the following resource types:

Azure > AI Foundry > RAI Policy

Policies

The following policies can be used to configure this control:

Azure > AI Foundry > RAI Policy > Content Filters

This control type relies on these other policies when running actions:

Permissions

Cloud permissions used by this control and its actions:

microsoft.cognitiveservices/accounts/raipolicies/write
microsoft.cognitiveservices/accounts/raipolicies/read

In Your Workspace

Developers

Control Type URI

tmod:@turbot/azure-aifoundry#/control/types/raiPolicyContentFilters

Category URI

tmod:@turbot/turbot#/control/categories/security

GraphQL

query controlType(id: "tmod:@turbot/azure-aifoundry#/control/types/raiPolicyContentFilters") { … }

query controls(filter: "controlTypeId:'tmod:@turbot/azure-aifoundry#/control/types/raiPolicyContentFilters'") { … }

CLI

Get Controls

turbot graphql controls --filter "controlTypeId:tmod:@turbot/azure-aifoundry#/control/types/raiPolicyContentFilters"