AIF-C01 Practice Question 10: Chatbot Safeguards PII
AWS Certified AI Practitioner (AIF-C01) · Test 1 · question 10 of 10
Which AWS capability should a bank use to stop its Amazon Bedrock customer chatbot from giving investment advice, and to mask account numbers and other personal data in both user messages and model responses?
Show answer and explanation
Answer:
Amazon Bedrock Guardrails evaluates both user inputs and model responses at inference time. A denied topic blocks subjects that are off-limits for the application, such as investment advice. Sensitive information filters block or mask personally identifiable information (PII) and custom patterns, such as account numbers. Guardrails also offers content filters for harmful content and prompt attacks, and contextual grounding checks for hallucinations (option A).
- B is wrong because Amazon Macie discovers sensitive data stored in Amazon S3. It doesn't filter live chatbot conversations or block topics. - C is wrong because AWS KMS protects data with encryption. Encryption doesn't stop the model from giving advice, and it doesn't mask PII in the text a user sees. - D is wrong because AWS CloudTrail records API activity for auditing. It logs what happened but doesn't block or redact anything.
Reference: Detect and filter harmful content by using Amazon Bedrock Guardrails