A company is implementing safety measures for a publicly available chatbot using Bedrock features. Which feature filters and blocks harmful content, disallowed topics, and confidential information at both input and output?

1 / 1
Select an answer
CorrectD

Explanation

Question Overview

Identify the Bedrock feature that blocks harmful output.

Requirements to satisfy
  • 1「harmful content, disallowed topics, and confidential information」Content that must not be produced
  • 2「filters and blocks」Safety filter = Amazon Bedrock Guardrails
Per-option explanation
AIncorrect

Amazon Bedrock Agents

This is incorrect. Amazon Bedrock Agents is a feature that breaks a user's request into multiple steps and executes them autonomously in sequence, calling external APIs or internal data (knowledge bases) as needed to complete the task. For example, a request such as 'Book a business trip' is planned and executed as a series of steps: search for availability → call booking API → send a confirmation email. It is not a safety filter for blocking harmful input and output.

BIncorrect

Amazon Bedrock Knowledge Bases

This is incorrect. Knowledge Bases is a feature for searching company documents and using them as the basis for answers (RAG). It does not block harmful input and output.

CIncorrect

Amazon Bedrock Model Evaluation

This is incorrect. Model Evaluation is a feature for measuring and comparing model quality. It is not a runtime input/output filter.

DCorrect

Amazon Bedrock Guardrails

This is correct. Bedrock Guardrails is a feature that filters and blocks harmful content, disallowed topics, and confidential information (PII) at both input and output, keeping responses safe and aligned with policy.

Key Takeaway

Amazon Bedrock Guardrails:
- Filters and blocks harmful content, disallowed topics, and confidential information (PII) at both input and output.
- Keeps responses safe and aligned with policy, supporting responsible use.

Related Links