Microsoft.AzureAI - AI Foundry Resources (not just project in Studio)

timrobertsusa 6 Reputation points
2026-01-22T00:19:03.1966667+00:00

Unable to Register Microsoft.AzureAI so I can use Prompt Sheild in AI Foundry via Code endpoints. Co-Pilot indicates that MS Support needs to enable this on my tennant/subscription. Create a support ticket does not filter me to request this to be done. I am unable to use code to evaluate and use Prompt Shield with the proper URI and Endpoint. Requirement: Tenant must be enabled for Foundry, Makes Microsoft.AzureAI provider visible. Until all this is done, I will never see the Foundry resource type and will only get the services.ai.azure.com Project endpoint.

Developer technologies | C#
Developer technologies | C#
An object-oriented and type-safe programming language that has its roots in the C family of languages and includes support for component-oriented programming.
{count} votes

1 answer

Sort by: Most helpful
  1. Surya Amrutha Vaishnavi Lanka (INFOSYS LIMITED) 1,725 Reputation points Microsoft External Staff
    2026-01-22T12:28:03.1533333+00:00

    Thanks for reaching out!

    Here are some steps which are helpful

    1. Raise ONE Support Ticket in Azure portal to the support team and fill them exactly in the category in issue type and technical option, Azure AI Foundry (If not visible, choose Other AI services) and give the problem type as Subscription / Tenant configuration
    2. You can paste the below message in the ticket " Please enable Azure AI Foundry (Microsoft.AzureAI provider) for my Azure tenant.  I am unable to register Microsoft.AzureAI and cannot access AI Foundry features via code endpoints such as Prompt Shield and Foundry project endpoints." This is a tenant-level enablement request.
    3. Submit the request and Wait for 1–3 working days Microsoft enables your tenant internally. No action needed from you during this time
    4. Verify after Microsoft Confirms, it should show as Microsoft.AzureAI — Registered.

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.