Amazon AWS Certified AI Practitioner AIF-C01 Practice Test 46
Question 450
The initial scenario was: A knowledge-grounded assistant encounters a document containing malicious instructions to leak system secrets. The original control is already in place. Now the assistant later needs an external transaction tool. Which ADDITIONAL action BEST addresses the new condition?
- Apply tool-level permission scoping and confirmation gates
- Red-team documents fail to trigger protected-data disclosure
- Treat any retrieved sentence beginning with 'system' as privileged
- Treat retrieved content as untrusted and enforce prompt-injection defenses
Explanation
Correct Answer
A:The assistant later needs an external transaction tool. The appropriate added action is: Apply tool-level permission scoping and confirmation gates.
Incorrect Answers
B:This describes evidence for the original approach, not a new action: Red-team documents fail to trigger protected-data disclosure.
C:This original shortcut remains unsuitable: External content does not gain instruction authority from its wording.
D:This was the original control and is already implemented. Retrieval text can carry adversarial instructions that must not override trusted policy.
Question 451
A public-facing Bedrock application needs to mask sensitive information in generated responses. Which option BEST meets the stated AI/ML or AWS requirement?
- Add storage encryption and governed retention controls
- Configure supported sensitive-information controls in Amazon Bedrock Guardrails
- Assume a single instruction to 'never reveal secrets' is full DLP
- Evaluate Bedrock AgentCore Identity or equivalent governed identity integration
Explanation
Correct Answer
B:Guardrails can screen or mask specified sensitive text patterns where configured. In this situation, choose: Configure supported sensitive-information controls in Amazon Bedrock Guardrails.
Incorrect Answers
A:This is appropriate only after a distinct change: Management later requires encryption of stored conversation logs. It is not the best initial response to a public-facing bedrock application needs to mask sensitive information in generated responses.
C:Prompt text alone does not guarantee protection against disclosure. The proposal, Assume a single instruction to 'never reveal secrets' is full DLP, does not satisfy this need.
D:This can be valid for a different task: A support agent must access third-party tools on behalf of individual authenticated employees with different permissions. The current request is different and needs the selected approach.
Question 452
A public-facing Bedrock application needs to mask sensitive information in generated responses. A colleague suggests: Assume a single instruction to 'never reveal secrets' is full DLP. Which finding BEST explains why the suggestion should be rejected?
- Shared broad credentials erase identity boundaries and increase risk
- Prompt text alone does not guarantee protection against disclosure
- The shortcut is safe because its terminology is familiar to the team
- Broad tool access creates an unauthorized financial action risk
Explanation
Correct Answer
B:Prompt text alone does not guarantee protection against disclosure. This is the direct defect in the proposed action: Assume a single instruction to 'never reveal secrets' is full DLP.
Incorrect Answers
A:This finding is associated with a different problem: A support agent must access third-party tools on behalf of individual authenticated employees with different permissions.
C:Familiar wording is not proof that an action is appropriate. The actual defect is: Prompt text alone does not guarantee protection against disclosure.
D:This concern belongs to another situation: An agent is permitted to read purchase orders but must not submit payments without authorization.
Question 453
A public-facing Bedrock application needs to mask sensitive information in generated responses. After implementing Configure supported sensitive-information controls in Amazon Bedrock Guardrails, which observation BEST verifies the intended result?
- The team schedules a future test of whether test prompts containing synthetic identifiers produce the prescribed safe handling, but has no result
- The team approves Configure supported sensitive-information controls in Amazon Bedrock Guardrails but does not run an operational test
- Logs correlate each authorized action with a distinct employee identity
- Test prompts containing synthetic identifiers produce the prescribed safe handling
Explanation
Correct Answer
D:The observed result, Test prompts containing synthetic identifiers produce the prescribed safe handling, provides direct evidence. Guardrails can screen or mask specified sensitive text patterns where configured.
Incorrect Answers
A:Verify the actual outcome: Test prompts containing synthetic identifiers produce the prescribed safe handling.
B:The required observable condition is: Test prompts containing synthetic identifiers produce the prescribed safe handling.
C:This observation would be evidence for another case: A support agent must access third-party tools on behalf of individual authenticated employees with different permissions, not the current scenario.
Question 454
The initial scenario was: A public-facing Bedrock application needs to mask sensitive information in generated responses. The original control is already in place. Now management later requires encryption of stored conversation logs. Which ADDITIONAL action BEST addresses the new condition?
- Add storage encryption and governed retention controls
- Assume a single instruction to 'never reveal secrets' is full DLP
- Test prompts containing synthetic identifiers produce the prescribed safe handling
- Configure supported sensitive-information controls in Amazon Bedrock Guardrails
Explanation
Correct Answer
A:Management later requires encryption of stored conversation logs. The appropriate added action is: Add storage encryption and governed retention controls.
Incorrect Answers
B:This original shortcut remains unsuitable: Prompt text alone does not guarantee protection against disclosure.
C:This describes evidence for the original approach, not a new action: Test prompts containing synthetic identifiers produce the prescribed safe handling.
D:This was the original control and is already implemented. Guardrails can screen or mask specified sensitive text patterns where configured.
Question 455
A support agent must access third-party tools on behalf of individual authenticated employees with different permissions. Which option BEST meets the stated AI/ML or AWS requirement?
- Give every user the same long-lived global API token
- Evaluate Bedrock AgentCore Identity or equivalent governed identity integration
- Apply action-level policy controls including AgentCore Policy where appropriate
- Add specific policy checks and tool-level authorization
Explanation
Correct Answer
B:Per-user identity and credential control supports scoped agent actions. In this situation, choose: Evaluate Bedrock AgentCore Identity or equivalent governed identity integration.
Incorrect Answers
A:Shared broad credentials erase identity boundaries and increase risk. The proposal, Give every user the same long-lived global API token, does not satisfy this need.
C:This can be valid for a different task: An agent is permitted to read purchase orders but must not submit payments without authorization. The current request is different and needs the selected approach.
D:This is appropriate only after a distinct change: Administrators later impose per-tool action restrictions. It is not the best initial response to a support agent must access third-party tools on behalf of individual authenticated employees with different permissions.
Question 456
A support agent must access third-party tools on behalf of individual authenticated employees with different permissions. A colleague suggests: Give every user the same long-lived global API token. 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
- Broad tool access creates an unauthorized financial action risk
- Shared broad credentials erase identity boundaries and increase risk
Explanation
Correct Answer
D:Shared broad credentials erase identity boundaries and increase risk. This is the direct defect in the proposed action: Give every user the same long-lived global API token.
Incorrect Answers
A:This concern belongs to another situation: 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: Shared broad credentials erase identity boundaries and increase risk.
C:This finding is associated with a different problem: An agent is permitted to read purchase orders but must not submit payments without authorization.
Question 457
A support agent must access third-party tools on behalf of individual authenticated employees with different permissions. After implementing Evaluate Bedrock AgentCore Identity or equivalent governed identity integration, which observation BEST verifies the intended result?
- The team schedules a future test of whether logs correlate each authorized action with a distinct employee identity, but has no result
- Tests confirm reads succeed while unapproved payments are denied
- Logs correlate each authorized action with a distinct employee identity
- The team approves Evaluate Bedrock AgentCore Identity or equivalent governed identity integration but does not run an operational test
Explanation
Correct Answer
C:The observed result, Logs correlate each authorized action with a distinct employee identity, provides direct evidence. Per-user identity and credential control supports scoped agent actions.
Incorrect Answers
A:An unperformed future check is not evidence. Verify the actual outcome: Logs correlate each authorized action with a distinct employee identity.
B:This observation would be evidence for another case: An agent is permitted to read purchase orders but must not submit payments without authorization, not the current scenario.
D:The required observable condition is: Logs correlate each authorized action with a distinct employee identity.
Question 458
The initial scenario was: A support agent must access third-party tools on behalf of individual authenticated employees with different permissions. The original control is already in place. Now administrators later impose per-tool action restrictions. Which ADDITIONAL action BEST addresses the new condition?
- Logs correlate each authorized action with a distinct employee identity
- Add specific policy checks and tool-level authorization
- Give every user the same long-lived global API token
- Evaluate Bedrock AgentCore Identity or equivalent governed identity integration
Explanation
Correct Answer
B:Administrators later impose per-tool action restrictions. The appropriate added action is: Add specific policy checks and tool-level authorization.
Incorrect Answers
A:This describes evidence for the original approach, not a new action: Logs correlate each authorized action with a distinct employee identity.
C:This original shortcut remains unsuitable: Shared broad credentials erase identity boundaries and increase risk.
D:This was the original control and is already implemented. Per-user identity and credential control supports scoped agent actions.
Question 459
An agent is permitted to read purchase orders but must not submit payments without authorization. Which option BEST meets the stated AI/ML or AWS requirement?
- Require a verified approval workflow before executing that write
- Ground outputs in approved RAG evidence and verify cited claims
- Apply action-level policy controls including AgentCore Policy where appropriate
- Give the agent unrestricted payment API access by default
Explanation
Correct Answer
C:Tool authorization must distinguish permitted reads from restricted writes. In this situation, choose: Apply action-level policy controls including AgentCore Policy where appropriate.
Incorrect Answers
A:This is appropriate only after a distinct change: A finance approver authorizes a specific one-time payment. It is not the best initial response to an agent is permitted to read purchase orders but must not submit payments without authorization.
B:This can be valid for a different task: An internal assistant answers compliance questions and must not present unsupported statements as established facts. The current request is different and needs the selected approach.
D:Broad tool access creates an unauthorized financial action risk. The proposal, Give the agent unrestricted payment API access by default, does not satisfy this need.
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.