updating per feedback from support

Author: Blackmist

articles/machine-learning/how-to-use-foundation-models.md

@@ -42,9 +42,12 @@ You can filter the list of models in the model catalog by Task, or by license. S
 You can quickly test out any pre-trained model using the Sample Inference widget on the model card, providing your own sample input to test the result. Additionally, the model card for each model includes a brief description of the model and links to samples for code based inferencing, finetuning and evaluation of the model.
 
 > [!IMPORTANT]
-> If you're using a private workspace, your virtual network needs to allow outbound access in order to use foundation models in Azure Machine Learning. 
+> Deploying foundational models to a managed online endpoint is currently supported with __public workspaces__ (and their public associated resources) only.
 >
-> Deploying foundational models to a managed online endpoint is currently supported with __public workspaces__ (and their public associated resources) only. When `egress_public_network_access` is set to `enabled` for a managed online endpoint deployment, the deployment can only access the resources with public access. For more information, see [Outbound resource access for managed online endpoints](how-to-secure-online-endpoint.md#outbound-resource-access).
+> * When `egress_public_network_access` is set to `disabled`, the deployment can only access the workspace-associated resources secured in the virtual network. 
+> * When `egress_public_network_access` is set to `enabled` for a managed online endpoint deployment, the deployment can only access the resources with public access. Which means that it cannot access resources secured in the virtual network.
+>
+> For more information, see [Outbound resource access for managed online endpoints](how-to-secure-online-endpoint.md#outbound-resource-access).
 
 ## How to evaluate foundation models using your own test data