AWS connect medium security documentation change
Summary
Restructured documentation with detailed steps for creating AI guardrails, added contextual grounding checks, sensitive info detection policies, testing recommendations, and publishing process
Security assessment
The changes document security-focused AI guardrail features including sensitive information detection, content filtering, and permission requirements. Explicitly mentions safeguards against hallucinations and data leaks, which are security controls.
Diff
diff --git a/connect/latest/adminguide/create-ai-guardrails.md b/connect/latest/adminguide/create-ai-guardrails.md index be8c5beb2..847ebb9ec 100644 --- a//connect/latest/adminguide/create-ai-guardrails.md +++ b//connect/latest/adminguide/create-ai-guardrails.md @@ -5 +5 @@ -Change the default blocked messageSample CLI commands +Important things to knowHow to create an AI guardrailChange the default blocked messageSample CLI commands @@ -9 +9,18 @@ Change the default blocked messageSample CLI commands -###### Important +An _AI guardrail_ is a resource that enables you to implement safeguards based on your use cases and responsible AI policies. + +Amazon Connect uses Amazon Bedrock guardrails. You can create and edit these guardrails in the Amazon Connect admin website. + +###### Contents + + * Important things to know + + * How to create an AI guardrail + + * Change the default blocked message + + * Sample CLI commands to configure AI guardrail policies + + + + +## Important things to know @@ -14,0 +32 @@ Change the default blocked messageSample CLI commands + * When configuring or editing a guardrail, we strongly recommend that you experiment and benchmark with different configurations. It's possible that some of your combinations may have unintended consequences. Test the guardrail to ensure that the results meet your use-case requirements. @@ -18 +35,0 @@ Change the default blocked messageSample CLI commands -An _AI guardrail_ is a resource that enables you to implement safeguards based on your use cases and responsible AI policies. @@ -20 +37 @@ An _AI guardrail_ is a resource that enables you to implement safeguards based o -Amazon Connect uses Amazon Bedrock guardrails. You can create and edit these guardrails in the Amazon Connect admin website. +## How to create an AI guardrail @@ -22 +39,9 @@ Amazon Connect uses Amazon Bedrock guardrails. You can create and edit these gua -Following is an overview of the policies that you can create and edit Amazon Connect admin website: + 1. Log in to the Amazon Connect admin website with an account that has **Amazon Q** , **AI guardrails - Create** permission in its security profile. + + 2. In the Amazon Connect admin website, on the left navigation menu, choose **Amazon Q** , **AI guardrails**. + + 3. On the **Guardrails** page, choose **Create Guardrail**. + + 4. On the **Create AI Guardrail** dialog box, enter a name and description of the guardrail, and then choose **Create**. + + 5. On the **AI Guardrail builder** page, complete the following fields as needed to create policies for your guardrail: @@ -27,0 +53,2 @@ Following is an overview of the policies that you can create and edit Amazon Con + * **Contextual grounding check** : Help detect and filter hallucinations in model responses based on grounding in a source and relevance to the user query. + @@ -34,2 +60,0 @@ Blocking or masking is done based on probabilistic detection of sensitive inform - * **Contextual grounding check** : Help detect and filter hallucinations in model responses based on grounding in a source and relevance to the user query. - @@ -37,0 +63 @@ Blocking or masking is done based on probabilistic detection of sensitive inform +Amazon Connect does not support **Image content filter** to help detect and filter inappropriate or toxic image content. @@ -38,0 +65 @@ Blocking or masking is done based on probabilistic detection of sensitive inform + 6. When your guardrail is complete, choose **Save**. @@ -39,0 +67 @@ Blocking or masking is done based on probabilistic detection of sensitive inform +When selecting from the versions dropdown, **Latest:Draft** always returns the saved state of the AI guardrail. @@ -41 +69,5 @@ Blocking or masking is done based on probabilistic detection of sensitive inform -Amazon Connect does not support **Image content filter** to help detect and filter inappropriate or toxic image content. + 7. Choose **Publish**. Updates to the AI guardrail are saved, the AI guardrail Visibility status is set to **Published** , and a new AI Guardrail version is created. + + + +When selecting from the versions dropdown, **Latest:Published** always returns the saved state of the AI guardrail. @@ -43 +74,0 @@ Amazon Connect does not support **Image content filter** to help detect and filt -###### Important @@ -45 +75,0 @@ Amazon Connect does not support **Image content filter** to help detect and filt -When configuring or editing a guardrail, we strongly recommend that you experiment and benchmark with different configurations. It's possible that some of your combinations may have unintended consequences. Test the guardrail to ensure that the results meet your use-case requirements. @@ -47 +76,0 @@ When configuring or editing a guardrail, we strongly recommend that you experime -The following section explains how to access the AI guardrail builder and editor in the Amazon Connect admin website, using the example of changing the blocked message that is displayed to users. @@ -50,0 +80,2 @@ The following section explains how to access the AI guardrail builder and editor +This section explains how to access the AI guardrail builder and editor in the Amazon Connect admin website, using the example of changing the blocked message that is displayed to users. +