AWS bedrock documentation change
Summary
Added console-based configuration for contextual grounding checks, including grounding and relevance thresholds
Security assessment
The change adds documentation about enabling grounding checks (verifying responses match source material) and relevance checks (ensuring responses stay on-topic). While these are security-adjacent controls to prevent hallucinations/misinformation, there's no evidence they address a specific vulnerability. The documentation focuses on general content filtering capabilities.
Diff
diff --git a/bedrock/latest/userguide/guardrails-contextual-grounding-check.md b/bedrock/latest/userguide/guardrails-contextual-grounding-check.md index 8b8a74eb1..f55cb5736 100644 --- a//bedrock/latest/userguide/guardrails-contextual-grounding-check.md +++ b//bedrock/latest/userguide/guardrails-contextual-grounding-check.md @@ -5 +5 @@ -Calling contextual grounding check with Invoke APIsCalling contextual grounding check with Converse APIsCalling contextual grounding check with ApplyGuardrail API +Add contextual grounding checks with the consoleCalling contextual grounding check with Invoke APIsCalling contextual grounding check with Converse APIsCalling contextual grounding check with ApplyGuardrail API @@ -144,0 +145,2 @@ Based on the above source, there can be four scenarios depending on the user’s + * Add contextual grounding checks with the console + @@ -153,0 +156,39 @@ Based on the above source, there can be four scenarios depending on the user’s +## Add contextual grounding checks with the console + + 1. Sign in to the AWS Management Console using an [IAM role with Amazon Bedrock permissions](./getting-started.html), and open the Amazon Bedrock console at [https://console.aws.amazon.com/bedrock/](https://console.aws.amazon.com/bedrock/). + + 2. From the left navigation pane, choose **Guardrails** , and then choose **Create guardrail**. + + 3. For **Provide guardrail details** page, do the following: + + 1. In the **Guardrail details** section, provide a **Name** and optional **Description** for the guardrail. + + 2. For **Messaging for blocked prompts** , enter a message that displays when your guardrail is applied. Select the **Apply the same blocked message for responses** checkbox to use the same message when your guardrail is applied on the response. + + 3. (Optional) To enable [cross-Region inference](./guardrails-cross-region.html) for your guardrail, expand **Cross-Region inference** , and then select **Enable cross-Region inference for your guardrail**. Choose a guardrail profile that defines the destination AWS Regions where guardrail inference requests can be routed. + + 4. (Optional) By default, your guardrail is encrypted with an AWS managed key. To use your own customer-managed KMS key, expand **KMS key selection** and select the **Customize encryption settings (advanced)** checkbox. + +You can select an existing AWS KMS key or select **Create an AWS KMS key** to create a new one. + + 5. (Optional) To add tags to your guardrail, expand **Tags** , and then, select **Add new tag** for each tag you define. + +For more information, see [Tagging Amazon Bedrock resources](./tagging.html). + + 6. Choose **Next**. + + 4. On the **Add contextual grounding check** page, configure thresholds to block un-grounded or irrelevant information. + +###### Note + +For each type of check, you can move the slider or input a threshold value from 0 to 0.99. Select an appropriate threshold for your uses. A higher threshold requires responses to be grounded or relevant with a high degree of confidence to be allowed. Responses below the threshold will be filtered. + + 1. In the **Grounding** field, select **Enable grounding check** to check if model responses are grounded. + + 2. In the **Relevance** field, select **Enable relevance check** to check if model responses are relevant. + + 3. When you finish configuring sensitive information filters, select **Next** or **Skip to review and create**. + + + + @@ -249 +290 @@ To use the Amazon Web Services Documentation, Javascript must be enabled. Please -Word filters +Add sensitive information filters