Amazon AWS Certified AI Practitioner AIF-C01 Practice Test 47
Question 460
An agent is permitted to read purchase orders but must not submit payments without authorization. A colleague suggests: Give the agent unrestricted payment API access by default. Which finding BEST explains why the suggestion should be rejected?
- Sampling diversity does not guarantee grounded accuracy
- The shortcut is safe because its terminology is familiar to the team
- Text output alone may omit actions performed by tools
- Broad tool access creates an unauthorized financial action risk
Explanation
Correct Answer
D:Broad tool access creates an unauthorized financial action risk. This is the direct defect in the proposed action: Give the agent unrestricted payment API access by default.
Incorrect Answers
A:This finding is associated with a different problem: An internal assistant answers compliance questions and must not present unsupported statements as established facts.
B:Familiar wording is not proof that an action is appropriate. The actual defect is: Broad tool access creates an unauthorized financial action risk.
C:This concern belongs to another situation: A security team needs an audit trail of an AI agent's tool invocations and identities for incident investigation.
Question 461
An agent is permitted to read purchase orders but must not submit payments without authorization. After implementing Apply action-level policy controls including AgentCore Policy where appropriate, which observation BEST verifies the intended result?
- Every sampled high-risk claim traces to an approved current source
- The team schedules a future test of whether tests confirm reads succeed while unapproved payments are denied, but has no result
- Tests confirm reads succeed while unapproved payments are denied
- The team approves Apply action-level policy controls including AgentCore Policy where appropriate but does not run an operational test
Explanation
Correct Answer
C:The observed result, Tests confirm reads succeed while unapproved payments are denied, provides direct evidence. Tool authorization must distinguish permitted reads from restricted writes.
Incorrect Answers
A:This observation would be evidence for another case: An internal assistant answers compliance questions and must not present unsupported statements as established facts, not the current scenario.
B:An unperformed future check is not evidence. Verify the actual outcome: Tests confirm reads succeed while unapproved payments are denied.
D:The required observable condition is: Tests confirm reads succeed while unapproved payments are denied.
Question 462
The initial scenario was: An agent is permitted to read purchase orders but must not submit payments without authorization. The original control is already in place. Now a finance approver authorizes a specific one-time payment. Which ADDITIONAL action BEST addresses the new condition?
- Tests confirm reads succeed while unapproved payments are denied
- Require a verified approval workflow before executing that write
- Give the agent unrestricted payment API access by default
- Apply action-level policy controls including AgentCore Policy where appropriate
Explanation
Correct Answer
B:A finance approver authorizes a specific one-time payment. The appropriate added action is: Require a verified approval workflow before executing that write.
Incorrect Answers
A:This describes evidence for the original approach, not a new action: Tests confirm reads succeed while unapproved payments are denied.
C:This original shortcut remains unsuitable: Broad tool access creates an unauthorized financial action risk.
D:This was the original control and is already implemented. Tool authorization must distinguish permitted reads from restricted writes.
Question 463
An internal assistant answers compliance questions and must not present unsupported statements as established facts. Which option BEST meets the stated AI/ML or AWS requirement?
- Capture governed logs of agent activity, identities and external tool calls
- Refresh the retrieval index and revoke outdated source access
- Ground outputs in approved RAG evidence and verify cited claims
- Rely on a higher temperature as proof all claims become factual
Explanation
Correct Answer
C:Retrieval plus validation can mitigate unsupported hallucinated statements. In this situation, choose: Ground outputs in approved RAG evidence and verify cited claims.
Incorrect Answers
A:This can be valid for a different task: A security team needs an audit trail of an AI agent's tool invocations and identities for incident investigation. The current request is different and needs the selected approach.
B:This is appropriate only after a distinct change: A source document is discovered to be obsolete. It is not the best initial response to an internal assistant answers compliance questions and must not present unsupported statements as established facts.
D:Sampling diversity does not guarantee grounded accuracy. The proposal, Rely on a higher temperature as proof all claims become factual, does not satisfy this need.
Question 464
An internal assistant answers compliance questions and must not present unsupported statements as established facts. A colleague suggests: Rely on a higher temperature as proof all claims become factual. Which finding BEST explains why the suggestion should be rejected?
- The shortcut is safe because its terminology is familiar to the team
- Text output alone may omit actions performed by tools
- Discarding lineage prevents reliable source attribution and audits
- Sampling diversity does not guarantee grounded accuracy
Explanation
Correct Answer
D:Sampling diversity does not guarantee grounded accuracy. This is the direct defect in the proposed action: Rely on a higher temperature as proof all claims become factual.
Incorrect Answers
A:Familiar wording is not proof that an action is appropriate. The actual defect is: Sampling diversity does not guarantee grounded accuracy.
B:This finding is associated with a different problem: A security team needs an audit trail of an AI agent's tool invocations and identities for incident investigation.
C:This concern belongs to another situation: A model reviewer wants to know which raw document a retrieved answer came from and which transformations altered it.
Question 465
An internal assistant answers compliance questions and must not present unsupported statements as established facts. After implementing Ground outputs in approved RAG evidence and verify cited claims, which observation BEST verifies the intended result?
- The team schedules a future test of whether every sampled high-risk claim traces to an approved current source, but has no result
- Authorized investigators can connect a tool call to an identity and timestamp
- The team approves Ground outputs in approved RAG evidence and verify cited claims but does not run an operational test
- Every sampled high-risk claim traces to an approved current source
Explanation
Correct Answer
D:The observed result, Every sampled high-risk claim traces to an approved current source, provides direct evidence. Retrieval plus validation can mitigate unsupported hallucinated statements.
Incorrect Answers
A:An unperformed future check is not evidence. Verify the actual outcome: Every sampled high-risk claim traces to an approved current source.
B:This observation would be evidence for another case: A security team needs an audit trail of an AI agent's tool invocations and identities for incident investigation, not the current scenario.
C:The required observable condition is: Every sampled high-risk claim traces to an approved current source.
Question 466
The initial scenario was: An internal assistant answers compliance questions and must not present unsupported statements as established facts. The original control is already in place. Now a source document is discovered to be obsolete. Which ADDITIONAL action BEST addresses the new condition?
- Every sampled high-risk claim traces to an approved current source
- Ground outputs in approved RAG evidence and verify cited claims
- Rely on a higher temperature as proof all claims become factual
- Refresh the retrieval index and revoke outdated source access
Explanation
Correct Answer
D:A source document is discovered to be obsolete. The appropriate added action is: Refresh the retrieval index and revoke outdated source access.
Incorrect Answers
A:This describes evidence for the original approach, not a new action: Every sampled high-risk claim traces to an approved current source.
B:This was the original control and is already implemented. Retrieval plus validation can mitigate unsupported hallucinated statements.
C:This original shortcut remains unsuitable: Sampling diversity does not guarantee grounded accuracy. It is not a valid solution to the changed requirement.
Question 467
A security team needs an audit trail of an AI agent's tool invocations and identities for incident investigation. Which option BEST meets the stated AI/ML or AWS requirement?
- Correlate activity traces with configuration and identity audit events
- Capture governed logs of agent activity, identities and external tool calls
- Record only the final conversational message and discard action traces
- Preserve data lineage and source identifiers across AI processing
Explanation
Correct Answer
B:Traceability enables accountability and investigation of AI-assisted actions. In this situation, choose: Capture governed logs of agent activity, identities and external tool calls.
Incorrect Answers
A:This is appropriate only after a distinct change: The team identifies suspicious changes in agent permissions. It is not the best initial response to a security team needs an audit trail of an ai agent's tool invocations and identities for incident investigation.
C:Text output alone may omit actions performed by tools. The proposal, Record only the final conversational message and discard action traces, does not satisfy this need.
D:This can be valid for a different task: A model reviewer wants to know which raw document a retrieved answer came from and which transformations altered it. The current request is different and needs the selected approach.
Question 468
A security team needs an audit trail of an AI agent's tool invocations and identities for incident investigation. A colleague suggests: Record only the final conversational message and discard action traces. Which finding BEST explains why the suggestion should be rejected?
- Discarding lineage prevents reliable source attribution and audits
- Text output alone may omit actions performed by tools
- Excessive permissions increase impact from compromised credentials or tools
- The shortcut is safe because its terminology is familiar to the team
Explanation
Correct Answer
B:Text output alone may omit actions performed by tools. This is the direct defect in the proposed action: Record only the final conversational message and discard action traces.
Incorrect Answers
A:This finding is associated with a different problem: A model reviewer wants to know which raw document a retrieved answer came from and which transformations altered it.
C:This concern belongs to another situation: A Bedrock-powered assistant needs read access to one S3 bucket and no privilege to modify unrelated AWS resources.
D:Familiar wording is not proof that an action is appropriate. The actual defect is: Text output alone may omit actions performed by tools.
Question 469
A security team needs an audit trail of an AI agent's tool invocations and identities for incident investigation. After implementing Capture governed logs of agent activity, identities and external tool calls, which observation BEST verifies the intended result?
- The team schedules a future test of whether authorized investigators can connect a tool call to an identity and timestamp, but has no result
- Authorized investigators can connect a tool call to an identity and timestamp
- The team approves Capture governed logs of agent activity, identities and external tool calls but does not run an operational test
- A sample answer is traced to its approved source and transform history
Explanation
Correct Answer
B:The observed result, Authorized investigators can connect a tool call to an identity and timestamp, provides direct evidence. Traceability enables accountability and investigation of AI-assisted actions.
Incorrect Answers
A:Verify the actual outcome: Authorized investigators can connect a tool call to an identity and timestamp.
C:The required observable condition is: Authorized investigators can connect a tool call to an identity and timestamp.
D:This observation would be evidence for another case: A model reviewer wants to know which raw document a retrieved answer came from and which transformations altered it, not the current scenario.
Question 470
The initial scenario was: A security team needs an audit trail of an AI agent's tool invocations and identities for incident investigation. The original control is already in place. Now the team identifies suspicious changes in agent permissions. Which ADDITIONAL action BEST addresses the new condition?
- Record only the final conversational message and discard action traces
- Correlate activity traces with configuration and identity audit events
- Authorized investigators can connect a tool call to an identity and timestamp
- Capture governed logs of agent activity, identities and external tool calls
Explanation
Correct Answer
B:The team identifies suspicious changes in agent permissions. The appropriate added action is: Correlate activity traces with configuration and identity audit events.
Incorrect Answers
A:This original shortcut remains unsuitable: Text output alone may omit actions performed by tools. It is not a valid solution to the changed requirement.
C:This describes evidence for the original approach, not a new action: Authorized investigators can connect a tool call to an identity and timestamp.
D:This was the original control and is already implemented. Traceability enables accountability and investigation of AI-assisted actions.