AWS bedrock documentation change
Summary
Improved documentation formatting (added colon, code formatting for API names), clarified policy configuration impacts on handling violations, and updated title from 'Stop' to 'Detect and filter' harmful content
Security assessment
Changes clarify existing security features (content filtering/masking) but do not indicate a specific security vulnerability being addressed. The title change reflects enhanced detection capabilities but doesn't reveal new security issues.
Diff
diff --git a/bedrock/latest/userguide/guardrails-how.md b/bedrock/latest/userguide/guardrails-how.md index b7b288d30..5803b8efc 100644 --- a//bedrock/latest/userguide/guardrails-how.md +++ b//bedrock/latest/userguide/guardrails-how.md @@ -11 +11 @@ Amazon Bedrock Guardrails helps keep your generative AI applications safe by eva -You can configure guardrails for your applications based on the following considerations +You can configure guardrails for your applications based on the following considerations: @@ -26 +26 @@ You can configure guardrails for your applications based on the following consid -When using a guardrail in the InvokeModel, InvokeModelWithResponseStream, Converse, or ConverseStream APIs, it works as follows during the inference call: +When using a guardrail in the `InvokeModel`, `InvokeModelWithResponseStream`, `Converse`, or `ConverseStream` operations, it works as follows during the inference call. (How this works depends on how you configure your policies to handle inputs and outputs.) @@ -34 +34 @@ When using a guardrail in the InvokeModel, InvokeModelWithResponseStream, Conver - * If the response results in a guardrail intervention or violation, it will be overridden with _pre-configured blocked messaging_ or _masking_ of the sensitive information. + * If the response results in a guardrail intervention or violation, it will be overridden with _pre-configured blocked messaging_ or _masking_ of the sensitive information based on your policy configuration. @@ -62 +62 @@ To use the Amazon Web Services Documentation, Javascript must be enabled. Please -Amazon Bedrock Guardrails: Stop harmful content +Amazon Bedrock Guardrails: Detect and filter harmful content