Add overview for Content Filtering and Guardrails

Author: brown9804

0_Azure/3_AzureAI/AIFoundry/demos/9_Content_Filtering.md

@@ -0,0 +1,60 @@
+# Content Filtering \& Guardrails - Overview
+
+Costa Rica
+
+[![GitHub](https://img.shields.io/badge/--181717?logo=github&logoColor=ffffff)](https://github.com/)
+[brown9804](https://github.com/brown9804)
+
+Last updated: 2025-12-26
+
+------------------------------------------
+
+> Microsoft Foundry enforces Responsible AI principles by applying content filters to all Large Language Models (LLMs) and image generation models.
+
+<details>
+<summary><b>List of References</b> (Click to expand)</summary>
+
+- [Content filtering overview](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/content-filter?view=foundry-classic)
+- [Configure content filters](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/content-filters?view=foundry-classic#understand-content-filter-configurability)
+- [Content filtering in Microsoft Foundry portal](https://learn.microsoft.com/en-us/azure/ai-foundry/concepts/content-filtering?view=foundry-classic)
+- [Azure OpenAI in Microsoft Foundry model deprecations and retirements](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/model-retirements?view=foundry-classic&tabs=text#current-models)
+- [Emissions Impact Dashboard for Azure](https://marketplace.microsoft.com/en-us/product/power-bi/coi-sustainability.emissions_impact_dashboard)
+- [GPT‑5.1 in Foundry: A Workhorse for Reasoning, Coding, and Chat](https://techcommunity.microsoft.com/blog/partnernews/gpt%E2%80%915-1-in-foundry-a-workhorse-for-reasoning-coding-and-chat/4469803)
+- [Getting started with Azure OpenAI Assistants (Preview)](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/assistant?view=foundry-classic)
+  
+</details>
+
+> [!NOTE]
+> These filters are powered by Azure AI Content Safety and operate at two stages:
+> - Input filtering: Checks user prompts before sending them to the model.
+> - Output filtering: Checks model responses before returning them to the user.
+
+> Default Filters:
+> - Every deployment has a `default filter applied automatically.`
+> - These filters `detect harmful content in four categories:`
+>   - Violence
+>   - Hate
+>   - Sexual
+>   - Self-harm
+> - Each category has severity levels: safe, low, medium, high.
+> - By default, medium and high severity content is blocked, while safe and low are allowed.
+
+## Custom Filters
+
+> You can create custom content filters to override defaults:
+> - Configure severity thresholds per category.
+> - Enable or disable binary classifiers (e.g., jailbreak detection).
+> - Custom filters can be applied at resource level or deployment level using:
+>   - Foundry portal
+>   - Azure CLI
+>   - Bicep templates
+
+> [!IMPORTANT]
+> By default, user's, any organizational data that flows through those agents is not shared publicly and is not used to train other models.
+
+<!-- START BADGE -->
+<div align="center">
+  <img src="https://img.shields.io/badge/Total%20views-1633-limegreen" alt="Total views">
+  <p>Refresh Date: 2025-12-03</p>
+</div>
+<!-- END BADGE -->