Pass Your CompTIA CAS-005 Exam Easy!

The reported condition calls for this action: Use the Diamond Model to structure those linked observations.

The current facts call for this action: use the Diamond Model to structure those linked observations.

Attack patterns organize recurring exploitation behaviors. The scenario does not present the alternate condition: a design team wants reusable examples of how attackers abuse password reset workflows. The proper immediate action is to use the Diamond Model to structure those linked observations.

The current facts call for this action: enforce separation of untrusted retrieval text from privileged instructions.

Data-sourced instructions should not trigger exfiltration. The scenario does not present the alternate condition: an assistant summarizes vendor notes that contain an instruction to email secret keys. The proper immediate action is to enforce separation of untrusted retrieval text from privileged instructions.

The reported condition calls for this action: Enforce separation of untrusted retrieval text from privileged instructions.

Rate-limit and test for data reconstruction or inversion behavior would not protect training provenance, write approvals and integrity monitoring.

Model pipelines have software supply-chain risks. The scenario does not present the alternate condition: a model provider distributes a new binary artifact with no signature or origin evidence. The proper immediate action is to protect training provenance, write approvals and integrity monitoring.

Removing unauthorized model export and enforcing artifact access controls would not protect training provenance, write approvals and integrity monitoring.

Requiring tested controls, documented datasets and measurable governance evidence would not restrict sensitive prompts and supply a governed private alternative.

The current facts call for this action: restrict sensitive prompts and supply a governed private alternative.

Requiring explainability, review and documented human appeal processes would not restrict sensitive prompts and supply a governed private alternative.

The reported condition calls for this action: Require policy-constrained approvals for high-impact changes.

The current facts call for this action: require policy-constrained approvals for high-impact changes.

AI identities still require resource-specific authorization. The scenario does not present the alternate condition: an agent's service identity has access to every customer record though it reads only order status. The proper immediate action is to require policy-constrained approvals for high-impact changes.

The current facts call for this action: confirm the instruction via a separate trusted authorization channel.

Biased detection can impede legitimate service. The scenario does not present the alternate condition: a fraud model flags every translated message as malicious without contextual evidence. The proper immediate action is to confirm the instruction via a separate trusted authorization channel.

The reported condition calls for this action: Confirm the instruction via a separate trusted authorization channel.

Applying bounded tool budgets and cancellation thresholds would not meet the need for least-privilege scopes and audited data access.

Business convenience cannot override mandatory data controls. The scenario does not present the alternate condition: an employee asks a work agent to bypass DLP so it can complete a report faster. The proper immediate action is to use least-privilege scopes and audited data access.

Requiring policy-constrained approvals for high-impact changes would not meet the need for least-privilege scopes and audited data access.

Confirming the instruction via a separate trusted authorization channel would not meet the need for protected transaction verification and behavioral reporting drills.

The current facts call for this action: use protected transaction verification and behavioral reporting drills.

Using strong out-of-band identity proofing rather than voice familiarity would not meet the need for protected transaction verification and behavioral reporting drills.

The reported condition calls for this action: Require explainability, review and documented human appeal processes.

The current facts call for this action: require explainability, review and documented human appeal processes.

Input disclosure may violate data obligations. The scenario does not present the alternate condition: employees paste customer medical histories into an unapproved public AI tool. The proper immediate action is to require explainability, review and documented human appeal processes.

The current facts call for this action: require signed provenance and dependency review before deployment.

Model outputs can leak memorized sensitive information. The scenario does not present the alternate condition: repeated crafted queries appear to reconstruct confidential training records. The proper immediate action is to require signed provenance and dependency review before deployment.

The reported condition calls for this action: Require signed provenance and dependency review before deployment.

Validating generated tool arguments against policy would not address the need for constrain tool permissions and require approval for sensitive actions.

Document content must not gain control authority. The scenario does not present the alternate condition: a retrieval assistant treats a sentence in an external document as an administrator command. The proper immediate action is to constrain tool permissions and require approval for sensitive actions.

Reduce secret-bearing prompts and test output disclosure controls would not address the need for constrain tool permissions and require approval for sensitive actions.