AIF-C01 Guidelines for Responsible AI Practice Question

A financial services company uses Amazon Bedrock to power a customer-facing chatbot that answers questions about loan products. During a compliance review, auditors ask the team to demonstrate that the chatbot's responses are grounded in approved policy documents and that the model is not generating unsupported financial advice. Which AWS service or feature should the team use to trace each response back to the specific source passages used to generate it?

⚠ Common exam trap

The trap here is assuming that contextual grounding checks in Amazon Bedrock Guardrails produce source citations, when they only detect and filter ungrounded responses without returning passage-level references.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Amazon Bedrock Knowledge Bases with citations enabled

The requirement is per-response provenance: showing which approved source passages produced each answer. Amazon Bedrock Knowledge Bases with citations enabled returns references to the retrieved chunks alongside the generated text, giving auditors a traceable map. Guardrails can detect ungrounded output but does not cite sources, while Model Monitor and CloudTrail address operational drift and API activity rather than content traceability.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✓
Amazon Bedrock Knowledge Bases with citations enabled
Why this is correct
Amazon Bedrock Knowledge Bases supports returning citations that map generated content back to the specific chunks retrieved from the underlying data source. This gives auditors a traceable link from each response to the approved policy document passages. Enabling citations in the RetrieveAndGenerate API response satisfies the requirement to demonstrate grounding and provenance for each chatbot answer.
✗
Amazon Bedrock Guardrails with contextual grounding checks
Why it's wrong here
Guardrails with contextual grounding can detect and filter responses that are not supported by the provided source content, and it can block hallucinated answers. However, it does not produce a traceable citation map from each generated sentence back to the exact source passage. For an audit that requires per-response provenance, the team needs a retrieval mechanism that returns references, not just a pass/fail grounding filter.
✗
AWS CloudTrail with Bedrock data events logging
Why it's wrong here
CloudTrail records API activity such as who invoked a Bedrock model and when, which supports audit trails of access. It does not capture the semantic link between a generated response and the source passages that informed it. Logging API calls alone cannot demonstrate that answers are grounded in approved documents, so it falls short of the traceability requirement.
✗
Amazon SageMaker Model Monitor with data quality baseline
Why it's wrong here
SageMaker Model Monitor detects drift and anomalies in model inputs and outputs over time by comparing against a baseline. It is designed for operational monitoring of deployed models, not for tracing individual chatbot responses to source documents. It cannot provide the per-response provenance the auditors require, so it does not meet the compliance need described.

Go deeper

AIF-C01 Guidelines for Responsible AI Practice Question

Why each option matters

Option-by-option breakdown

Related to this question