Merge pull request #4730 from PatrickFarley/aoai-build

add spotlighting info

Author: AnnaMHuff

articles/ai-services/openai/concepts/content-filter-prompt-shields.md

@@ -15,4 +15,13 @@ Prompt shields are a feature of the Azure OpenAI content filtering system that h
 
 Prompt shields analyzes LLM inputs and detects adversarial user input attacks.
 
-[!INCLUDE [prompt-shield-attack-info](../../content-safety/includes/prompt-shield-attack-info.md)]
+[!INCLUDE [prompt-shield-attack-info](../../content-safety/includes/prompt-shield-attack-info.md)]
+
+## Spotlighting for prompt shields
+
+Spotlighting is a sub-feature of prompt shields that enhances protection against indirect attacks by tagging the input documents with special formatting to indicate lower trust to the model. When spotlighting is enabled, the service transforms the document content using base-64 encoding, and the model is configured to treat this content as less trustworthy than direct user and system prompts. This helps prevent the model from executing unintended commands or actions that are found in the content of the documents.
+
+Spotlighting is turned off by default, and users can enable it when [configuring their content filter](../how-to/content-filters.md) in the Azure AI Foundry portal or REST API.
+
+There is no direct cost for spotlighting, but it adds more tokens to a user's prompt, which could increase the total costs. Also note that spotlighting could make a lengthy document surpass the input size limit. 
+

articles/ai-services/openai/whats-new.md

@@ -27,7 +27,7 @@ Personally identifiable information (PII) detection is now available as a built-
 
 ### Spotlighting for prompt shields
 
-Spotlighting is a new feature for prompt shields that enhances protection from indirect (embedded document) attacks by tagging input documents with special formatting to indicate lower trust to the model. For more information, see the [Prompt shields content filtering](./concepts/content-filter-prompt-shields.md) documentation.
+Spotlighting is a sub-feature of prompt shields that enhances protection against indirect (embedded document) attacks by tagging input documents with special formatting to indicate lower trust to the model. For more information, see the [Prompt shields content filtering](./concepts/content-filter-prompt-shields.md) documentation.