An Azure service that enables users to identify content that is potentially offensive, risky, or otherwise undesirable. Previously known as Azure Content Moderator.
Azure AI Foundry Content Filter - "Annotate Only" option grayed out for Input Filter despite Global Admin permissions
George Mikael
0
Reputation points
Hello,
I'm attempting to create a custom content filter in Azure AI Foundry and have run into an issue with the Input Filter configuration.
Issue:
When creating a custom content filter under:
Guardrails & Controls → Content Filters → Create Content Filter
the Action field for the content categories (Violence, Hate, Sexual, and Self-Harm) is locked and only shows "Annotate and Block."
The "Annotate Only" option is grayed out and cannot be selected.
Environment:
- Azure AI Foundry / Azure OpenAI
- Custom Content Filter
- GPT deployment: GPT-5.6-Terra Subscription Type: EA
- User Role:
- Microsoft Entra Global Administrator
- Azure Subscription Owner
- Azure AI Foundry Project Owner
- Resource Type: Azure AI Services / Azure OpenAI
- Azure Subscription Owner
- Microsoft Entra Global Administrator
- User Role:
Questions:
- Is the inability to select Annotate Only expected behavior?
- Does this require approval for Modified Content Filters or a Limited Access Review?
- Is this feature dependent on:
- Azure subscription type?
- Region?
- Model family?
- Resource configuration?
- Has anyone successfully enabled Annotate Only for content categories in Azure AI Foundry, and if so, what prerequisite was required?
- Are there specific Azure RBAC roles beyond Global Administrator and Subscription Owner that are needed?
- Model family?
- Region?
- Azure subscription type?
Observations:
- The Action field is disabled rather than generating a permissions error.
- Other settings within the content filter are editable.
- The issue occurs regardless of which administrative account is used.
- I have not found documentation clearly stating why the control would be disabled.
Any guidance from others who have encountered this would be appreciated.
Thank you.
Content Safety in Foundry Control Plane
Content Safety in Foundry Control Plane
Sign in to answer