Merge pull request #53940 from MicrosoftDocs/main

learn-build-service-prod[bot] · web-flow · commit 482a86db11b9 · 2026-03-23T17:13:28.000Z
Auto Publish – main to live - 2026-03-23 17:00 UTC
diff --git a/learn-pr/wwl-azure/design-implement-hybrid-networking/includes/2-design-implement-vpn-gateway.md b/learn-pr/wwl-azure/design-implement-hybrid-networking/includes/2-design-implement-vpn-gateway.md
@@ -27,8 +27,7 @@ Creating a virtual network gateway can take some time to complete, so it's vital
 When you create a VPN Gateway, you must select the [gateway SKU](/azure/vpn-gateway/about-gateway-skus). You can select a SKU that satisfies your requirements based on the workload, throughput, feature, and SLAs. 
 
 > [!TIP]
-> The Basic SKU has feature and performance limitations and shouldn't be used in  production environments. 
-
+> The Basic SKU has feature and performance limitations and shouldn't be used in production environments. For new gateways, use availability zone (AZ) SKUs. 
 
 ## VPN Gateway types
 
@@ -77,4 +76,4 @@ You can create an Azure VPN gateway in an active-standby configuration. Active-s
 
 You can create an Azure VPN gateway in an active-active configuration. In the active-active configuration, both instances of the gateway establish VPN tunnels to your on-premises VPN device. The traffic from your Azure virtual network to your on-premises network is routed through both tunnels simultaneously. If an outage occurs, Azure automatically switches to the other tunnel. 
 
-:::image type="content" source="../media/vpn-active-active-89241ba7.png" alt-text="Diagram of an Azure VPN Gateway in active-active configuration.":::
+:::image type="content" source="../media/vpn-active-active-89241ba7.png" alt-text="Diagram of an Azure VPN Gateway in active-active configuration.":::
diff --git a/learn-pr/wwl-azure/design-implement-hybrid-networking/index.yml b/learn-pr/wwl-azure/design-implement-hybrid-networking/index.yml
@@ -5,7 +5,7 @@ metadata:
   prefetch-feature-rollout: true
   title: Design and Implement Hybrid Networking
   description: "You learn how to design and implement hybrid networking solutions such as Site-to-Site VPN connections, Point-to-Site VPN connections, Azure Virtual WAN, and Virtual WAN hubs."
-  ms.date: 09/09/2025
+  ms.date: 03/23/2026
   author: wwlpublish
   ms.author: cynthist
   ms.topic: module
@@ -56,4 +56,4 @@ units:
 - learn.wwl.design-implement-hybrid-networking.create-network-virtual-appliance-nva-virtual-hub
 - learn.wwl.design-implement-hybrid-networking.summary
 badge:
-  uid: learn.wwl.design-implement-hybrid-networking.badge
+  uid: learn.wwl.design-implement-hybrid-networking.badge
diff --git a/learn-pr/wwl-data-ai/foundry-sdk/07-knowledge-check.yml b/learn-pr/wwl-data-ai/foundry-sdk/07-knowledge-check.yml
@@ -15,39 +15,17 @@ durationInMinutes: 5
 quiz:
   title: Check your knowledge
   questions:
-    - content: What is the primary purpose of the project client in the Microsoft Foundry SDK?
+    - content: Which endpoint offers the broadest support for OpenAI APIs with Foundry Models?
       choices:
-        - content: To generate AI responses using the Responses API
+        - content: The Foundry project endpoint
           isCorrect: false
-          explanation: Incorrect. The OpenAI-compatible client is used to generate AI responses.
-        - content: To access Foundry-native operations like listing connections and managing project properties
+          explanation: Incorrect. The Foundry Project endpoint is used for Foundry-native operations, and OpenAI API support for the Responses API with Foundry direct models.
+        - content: The Azure OpenAI endpoint
           isCorrect: true
-          explanation: Correct. The project client provides access to Foundry-native operations that don't have OpenAI equivalents.
-        - content: To deploy new AI models to the project
+          explanation: Correct. The Azure OpenAI endpoint provides broad support for OpenAI APIs with Foundry Models.
+        - content: The Foundry Tools endpoint
           isCorrect: false
-          explanation: Incorrect. Model deployment is done through the Foundry portal, not the SDK.
-    - content: Which method do you use to generate responses with the Responses API?
-      choices:
-        - content: client.chat.completions.create()
-          isCorrect: false
-          explanation: Incorrect. This is the older chat completions API method.
-        - content: client.responses.create()
-          isCorrect: true
-          explanation: Correct. The Responses API uses the responses.create() method.
-        - content: client.generate.response()
-          isCorrect: false
-          explanation: Incorrect. This method doesn't exist in the SDK.
-    - content: What does the previous_response_id parameter do in the Responses API?
-      choices:
-        - content: It links responses together to maintain conversation context
-          isCorrect: true
-          explanation: Correct. The previous_response_id parameter maintains conversation context across multiple API calls.
-        - content: It retrieves an older response from the database
-          isCorrect: false
-          explanation: Incorrect. Use the responses.retrieve() method to retrieve previous responses.
-        - content: It deletes previous responses to save storage space
-          isCorrect: false
-          explanation: Incorrect. The parameter doesn't delete responses.
+          explanation: Incorrect. Foundry Tools do not provide broad support for OpenAI APIs with Foundry Models.
     - content: Which package must you install to use the Microsoft Foundry SDK in Python?
       choices:
         - content: Package `azure-foundry`
@@ -59,14 +37,14 @@ quiz:
         - content: Package `microsoft-foundry-sdk`
           isCorrect: false
           explanation: Incorrect. This package doesn't exist.
-    - content: What advantage does the Responses API offer over the ChatCompletions API?
+    - content: Which method do you use to generate responses with the Responses API?
       choices:
-        - content: It works only with Azure OpenAI models
+        - content: client.chat.completions.create()
           isCorrect: false
-          explanation: Incorrect. The Responses API works with both Azure OpenAI and Foundry direct models.
-        - content: It provides stateful, multi-turn conversation support
-          isCorrect: true
-          explanation: Correct. The Responses API maintains conversation context and provides stateful interactions.
-        - content: It requires less authentication configuration
+          explanation: Incorrect. This is the older chat completions API method.
+        - content: client.get_response_id()
           isCorrect: false
-          explanation: Incorrect. Both APIs require the same authentication approach.
+          explanation: Incorrect. This method doesn't generate a response.
+        - content: client.responses.create()
+          isCorrect: true
+          explanation: Correct. The Responses API uses the responses.create() method.
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/1-introduction.yml b/learn-pr/wwl-data-ai/responsible-ai-studio/1-introduction.yml
@@ -6,11 +6,10 @@ metadata:
   description: Introduce responsible generative AI.
   author: ivorb
   ms.author: berryivor
-  ms.date: 02/26/2026
+  ms.date: 03/22/2026
   ms.topic: unit
   ms.collection:
     - wwl-ai-copilot
 durationInMinutes: 1
 content: |
   [!include[](includes/1-introduction.md)]
-
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/2-plan-responsible-ai.yml b/learn-pr/wwl-data-ai/responsible-ai-studio/2-plan-responsible-ai.yml
@@ -6,11 +6,10 @@ metadata:
   description: Describe a practical process for responsible generative AI development.
   author: ivorb
   ms.author: berryivor
-  ms.date: 02/26/2026
+  ms.date: 03/22/2026
   ms.topic: unit
   ms.collection:
     - wwl-ai-copilot
 durationInMinutes: 2
 content: |
   [!include[](includes/2-plan-responsible-ai.md)]
-
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/3-identify-harms.yml b/learn-pr/wwl-data-ai/responsible-ai-studio/3-identify-harms.yml
@@ -6,11 +6,10 @@ metadata:
   description: Identify, prioritize, validate and document potential harms in a responsible AI solution.
   author: ivorb
   ms.author: berryivor
-  ms.date: 02/26/2026
+  ms.date: 03/22/2026
   ms.topic: unit
   ms.collection:
     - wwl-ai-copilot
 durationInMinutes: 5
 content: |
   [!include[](includes/3-identify-harms.md)]
-
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/4-measure-harms.yml b/learn-pr/wwl-data-ai/responsible-ai-studio/4-measure-harms.yml
@@ -6,11 +6,10 @@ metadata:
   description: Measure the presence of potential harm in a responsible AI solution.
   author: ivorb
   ms.author: berryivor
-  ms.date: 02/26/2026
+  ms.date: 03/22/2026
   ms.topic: unit
   ms.collection:
     - wwl-ai-copilot
 durationInMinutes: 5
 content: |
   [!include[](includes/4-measure-harms.md)]
-
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/5-mitigate-harms.yml b/learn-pr/wwl-data-ai/responsible-ai-studio/5-mitigate-harms.yml
@@ -6,11 +6,10 @@ metadata:
   description: Mitigate potential harms at multiple levels of a responsible AI solution.
   author: ivorb
   ms.author: berryivor
-  ms.date: 02/26/2026
+  ms.date: 03/22/2026
   ms.topic: unit
   ms.collection:
     - wwl-ai-copilot
 durationInMinutes: 5
 content: |
   [!include[](includes/5-mitigate-harms.md)]
-
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/6-operate-responsibly.yml b/learn-pr/wwl-data-ai/responsible-ai-studio/6-operate-responsibly.yml
@@ -6,11 +6,10 @@ metadata:
   description: Release and operate a generative AI solution responsibly.
   author: ivorb
   ms.author: berryivor
-  ms.date: 02/26/2026
+  ms.date: 03/22/2026
   ms.topic: unit
   ms.collection:
     - wwl-ai-copilot
 durationInMinutes: 3
 content: |
   [!include[](includes/6-operate-responsibly.md)]
-
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/7-exercise-content-filters.yml b/learn-pr/wwl-data-ai/responsible-ai-studio/7-exercise-content-filters.yml
@@ -1,17 +1,15 @@
 ### YamlMime:ModuleUnit
 uid: learn.wwl.responsible-ai-studio.exercise-content-filters
-title: Exercise - Apply content filters to prevent the output of harmful content
+title: Exercise - Apply guardrails to prevent the output of harmful content
 metadata:
-  title: Exercise - Apply content filters to prevent the output of harmful content
-  description: Hands-on exercise to explore content filters in Microsoft Foundry Portal.
+  title: Exercise - Apply guardrails to prevent the output of harmful content
+  description: Hands-on exercise to explore guardrails in Microsoft Foundry Portal.
   author: ivorb
   ms.author: berryivor
-  ms.date: 02/26/2026
+  ms.date: 03/22/2026
   ms.topic: unit
   ms.collection:
     - wwl-ai-copilot
-durationInMinutes: 25
+durationInMinutes: 20
 content: |
   [!include[](includes/7-exercise-content-filters.md)]
-
-
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/8-knowledge-check.yml b/learn-pr/wwl-data-ai/responsible-ai-studio/8-knowledge-check.yml
@@ -6,47 +6,45 @@ metadata:
   description: Check your understanding of the content presented in this module.
   author: ivorb
   ms.author: berryivor
-  ms.date: 02/26/2026
+  ms.date: 03/22/2026
   ms.topic: unit
   ms.collection:
     - wwl-ai-copilot
   module_assessment: true
 durationInMinutes: 3
 content: |
-quiz:  
-  questions:  
-  - content: "Why should you consider creating an AI Impact Assessment when designing a generative AI solution?"  
-    choices:  
-    - content: "To make a legal case that indemnifies you from responsibility for harms caused by the solution"  
-      isCorrect: false  
-      explanation: "An AI Impact Assessment does not indemnify you from responsibility for harm"  
-    - content: "To document the purpose, expected use, and potential harms for the solution"  
-      isCorrect: true  
-      explanation: "An AI Impact Assessment guide documents the expected use of the system and helps identify potential harms."  
-    - content: "To evaluate the cost of cloud services required to implement your solution"  
-      isCorrect: false  
-      explanation: "An AI Impact Assessment typically doesn't include evaluations of cloud service costs."  
-  - content: "What capability of Microsoft Foundry helps mitigate harmful content generation at the Safety System level?"  
-    choices:  
-    - content: "DALL-E model support"  
-      isCorrect: false  
-      explanation: "DALL-E models are used to generate images."  
-    - content: "Fine-tuning"  
-      isCorrect: false  
-      explanation: "Fine-tuning is used to customize models, and provides mitigation at the Model layer"  
-    - content: "Content filters"  
-      isCorrect: true  
-      explanation: "Content filters enable you to suppress harmful content at the Safety System layer."  
-  - content: "Why should you consider a phased delivery plan for your generative AI solution?"  
-    choices:  
-    - content: "To enable you to gather feedback and identify issues before releasing the solution more broadly"  
-      isCorrect: true  
-      explanation: "An initial release to a restricted user base enables you to minimize harm by gather feedback and identifying issues before broad release."  
-    - content: "To eliminate the need to map, measure, mitigate, and manage potential harms"  
-      isCorrect: false  
-      explanation: "A phased delivery plan doesn't eliminate the need to identify, measure, and mitigate potential harms."  
-    - content: "To enable you to charge more for the solution"  
-      isCorrect: false  
-      explanation: "The goal of a phased delivery plan is to reduce potential harm, not to maximize revenue."  
-
-
+quiz:
+  questions:
+    - content: "Why should you consider creating an AI Impact Assessment when designing a generative AI solution?"
+      choices:
+        - content: "To make a legal case that indemnifies you from responsibility for harms caused by the solution"
+          isCorrect: false
+          explanation: "An AI Impact Assessment does not indemnify you from responsibility for harm"
+        - content: "To document the purpose, expected use, and potential harms for the solution"
+          isCorrect: true
+          explanation: "An AI Impact Assessment guide documents the expected use of the system and helps identify potential harms."
+        - content: "To evaluate the cost of cloud services required to implement your solution"
+          isCorrect: false
+          explanation: "An AI Impact Assessment typically doesn't include evaluations of cloud service costs."
+    - content: "What capability of Microsoft Foundry helps mitigate harmful content generation at the Safety System level?"
+      choices:
+        - content: "DALL-E model support"
+          isCorrect: false
+          explanation: "DALL-E models are used to generate images."
+        - content: "Fine-tuning"
+          isCorrect: false
+          explanation: "Fine-tuning is used to customize models, and provides mitigation at the Model layer"
+        - content: "Guardrails"
+          isCorrect: true
+          explanation: "Guardrails enable you to suppress harmful content at the Safety System layer."
+    - content: "Why should you consider a phased delivery plan for your generative AI solution?"
+      choices:
+        - content: "To enable you to gather feedback and identify issues before releasing the solution more broadly"
+          isCorrect: true
+          explanation: "An initial release to a restricted user base enables you to minimize harm by gather feedback and identifying issues before broad release."
+        - content: "To eliminate the need to map, measure, mitigate, and manage potential harms"
+          isCorrect: false
+          explanation: "A phased delivery plan doesn't eliminate the need to identify, measure, and mitigate potential harms."
+        - content: "To enable you to charge more for the solution"
+          isCorrect: false
+          explanation: "The goal of a phased delivery plan is to reduce potential harm, not to maximize revenue."
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/9-summary.yml b/learn-pr/wwl-data-ai/responsible-ai-studio/9-summary.yml
@@ -6,11 +6,10 @@ metadata:
   description: Summarize a responsible approach for generative AI.
   author: ivorb
   ms.author: berryivor
-  ms.date: 02/26/2026
+  ms.date: 03/22/2026
   ms.topic: unit
   ms.collection:
     - wwl-ai-copilot
 durationInMinutes: 1
 content: |
   [!include[](includes/9-summary.md)]
-
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/includes/5-mitigate-harms.md b/learn-pr/wwl-data-ai/responsible-ai-studio/includes/5-mitigate-harms.md
@@ -15,14 +15,14 @@ The model layer consists of one or more generative AI models at the heart of you
 
 Mitigations you can apply at the model layer include:
 
-- Selecting a model that is appropriate for the intended solution use. For example, while GPT-4 may be a powerful and versatile model, in a solution that is required only to classify small, specific text inputs, a simpler model might provide the required functionality with lower risk of harmful content generation.
+- Selecting a model that's appropriate for the intended solution use. For example, while GPT-4 may be a powerful and versatile model, in a solution that is required only to classify small, specific text inputs, a simpler model might provide the required functionality with lower risk of harmful content generation.
 - *Fine-tuning* a foundational model with your own training data so that the responses it generates are more likely to be relevant and scoped to your solution scenario.
 
 ## 2: The *safety system* layer
 
-The safety system layer includes platform-level configurations and capabilities that help mitigate harm. For example, Microsoft Foundry includes support for *content filters* that apply criteria to suppress prompts and responses based on classification of content into four severity levels (*safe*, *low*, *medium*, and *high*) for four categories of potential harm (*hate*, *sexual*, *violence*, and *self-harm*).
+The safety system layer includes platform-level configurations and capabilities that help mitigate harm. For example, Microsoft Foundry includes support for *guardrails* that apply criteria to suppress prompts and responses based on *content filters* that classify content into four severity levels (*safe*, *low*, *medium*, and *high*) for five categories of potential harm (*hate and fairness*, *sexual*, *violence*, *self-harm*, and *task-adherence*).
 
-Other safety system layer mitigations can include abuse detection algorithms to determine if the solution is being systematically abused (for example through high volumes of automated requests from a bot) and alert notifications that enable a fast response to potential system abuse or harmful behavior.
+Other safety system layer mitigations in Foundry guardrails include *prompt shields* that use abuse detection algorithms to determine if the solution is being systematically abused (for example by a user attempting to subvert the system prompt).
 
 ## 3: The *system message and grounding* layer
 
@@ -39,4 +39,3 @@ The user experience layer includes the software application through which users
 Designing the application user interface to constrain inputs to specific subjects or types, or applying input and output validation can mitigate the risk of potentially harmful responses.
 
 Documentation and other descriptions of a generative AI solution should be appropriately transparent about the capabilities and limitations of the system, the models on which it's based, and any potential harms that may not always be addressed by the mitigation measures you have put in place.
-
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/includes/6-operate-responsibly.md b/learn-pr/wwl-data-ai/responsible-ai-studio/includes/6-operate-responsibly.md
diff --git a/learn-pr/wwl-data-ai/responsible-ai-studio/index.yml b/learn-pr/wwl-data-ai/responsible-ai-studio/index.yml
diff --git a/learn-pr/wwl-data-ai/use-generative-ai-tools/includes/05-file-search.md b/learn-pr/wwl-data-ai/use-generative-ai-tools/includes/05-file-search.md
