Support Entra to authenticate to Azure models

[Advitya17]

• Copilot Chat Extension Version:
• VS Code Version:
• OS Version:
• Feature (e.g. agent/edit/ask mode): Model Picker
• Selected model (e.g. GPT 4.1, Claude 3.7 Sonnet):
• Logs:

Steps to Reproduce:
If I try to add a custom fine-tuned Azure AI Foundry deployment from the Model Picker, the option asks me for an API key.

This is unfortunately not workable, as Key-based Auth is disabled for our entire org due to our latest security wave, and enabling it will create an incident asking me to undo it.

Instead of Key-based Auth support which we're moving away from, could we please add Entra ID based support when trying to add an Azure deployment?
Under the hood, the key/token can be grabbed from user's Entra ID with the "https://cognitiveservices.azure.com/.default" scope. And this may also remove the need of creating a toggle in .vscode/settings.json just to have VS Code utilize our FT model instead.

[Advitya17]

Hi @lramos15, could I please get an ETA on this task? We're blocked on this support to be able to use our FT model deployment (hosted on Azure AI Foundry), and have devs in our org looking to try this experience with our deployment. We're unable to use Key-based Auth due to the latest Security wave as discussed before.

[eleanorjboyd]

Hi @Advitya17, sorry for the delay here. I will investigate ETA and scope here and circle back- thanks