A company is creating a child-facing interactive app that generates new stories from classics using Amazon Bedrock. They need to ensure outputs and topics are child-appropriate. Which AWS capability should they use?

Guardrails for Amazon Bedrock is the correct choice because it allows you to define safety policies to filter out harmful content, ensuring that the generated stories and topics are child-appropriate. This is crucial for applications targeting young audiences, as it helps prevent the generation of inappropriate or unsafe content. Amazon Rekognition is an image and video analysis service, not designed for text-based content moderation in this context. Amazon Bedrock playgrounds are interactive environments for experimenting with models but don't inherently provide content moderation policies. Agents for Amazon Bedrock are used for orchestrating complex tasks and interacting with external systems, not primarily for content filtering.