100% Latest & Updated Palo Alto Networks NetSec-Architect Practice Test Questions, Exam Dumps & Verified Answers!
30 Days Free Updates, Instant Download!
Question 110
Which validation most directly demonstrates ai kubernetes east-west security is correctly designed for North America regional bank? After North America regional bank completed a limited security transformation pilot, reviewers observed a control gap: AI inference pods in Kubernetes share a trust domain and sensitive model-service calls bypass inspection. The case assumes the affected flows and dependencies have been traced and all unrelated systems are functioning; choose the narrow design repair before wider deployment.
- For North America regional bank: Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path.
- For North America regional bank: Attempt the same permitted business operation in an approved workspace and in a disallowed app.
- For North America regional bank: Call the test agent with benign and policy-violating prompts and observe API verdict handling.
- For North America regional bank: Audit a sample AI request from user identity to data classification, model path, and policy decision.
Explanation
Correct Answer
A:This checks the relevant ai kubernetes east-west security enforcement path rather than a distant symptom. Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path. The expected evidence is The intended workload flows reach runtime controls while unauthorized lateral calls fail. Case R152-C028; angle 2; post-pilot control gap; option considered: Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path.
Incorrect Answers
B:This option addresses a different design objective: Sanctioned AI app policy, not the primary control for ai kubernetes east-west security. It would be relevant where Two departments use different AI apps but security cannot distinguish sanctioned workspaces from unknown services. Here the necessary evidence is Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path. Case R152-C028; angle 2; post-pilot control gap; option considered: Attempt the same permitted business operation in an approved workspace and in a disallowed app.
C:This option addresses a different design objective: AI runtime API intercept, not the primary control for ai kubernetes east-west security. It would be relevant where A serverless AI agent needs prompt and response scanning but cannot be routed through a separate firewall hop. Here the necessary evidence is Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path. Case R152-C028; angle 2; post-pilot control gap; option considered: Call the test agent with benign and policy-violating prompts and observe API verdict handling.
D:This option addresses a different design objective: AI compliance evidence chain, not the primary control for ai kubernetes east-west security. It would be relevant where An organization handling regulated records proposes AI deployment without data-flow inventory or model risk evidence. Here the necessary evidence is Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path. Case R152-C028; angle 2; post-pilot control gap; option considered: Audit a sample AI request from user identity to data classification, model path, and policy decision.
Question 111
Which design assumption should the architect REJECT when evaluating ai kubernetes east-west security for North America regional bank? After North America regional bank completed a limited security transformation pilot, reviewers observed a control gap: AI inference pods in Kubernetes share a trust domain and sensitive model-service calls bypass inspection. The case assumes the affected flows and dependencies have been traced and all unrelated systems are functioning; choose the narrow design repair before wider deployment.
- For North America regional bank: The intended workload flows reach runtime controls while unauthorized lateral calls fail.
- For North America regional bank: Design a supported Prisma AIRS runtime placement and cluster segmentation to cover actual pod-to-model flows.
- For North America regional bank: Add an internet edge firewall that never observes the intra-cluster model traffic.
- For North America regional bank: Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path.
Explanation
Correct Answer
C:Reject this assumption: Add an internet edge firewall that never observes the intra-cluster model traffic. It does not establish ai kubernetes east-west security; the supported architecture is Design a supported Prisma AIRS runtime placement and cluster segmentation to cover actual pod-to-model flows. Verify it by: Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path. Case R152-C028; angle 3; post-pilot control gap; option considered: Add an internet edge firewall that never observes the intra-cluster model traffic.
Incorrect Answers
A:This is a valid design/control/evidence statement in the specified case, not the assumption to reject. The intended workload flows reach runtime controls while unauthorized lateral calls fail. It is consistent with the required outcome: The intended workload flows reach runtime controls while unauthorized lateral calls fail. Case R152-C028; angle 3; post-pilot control gap; option considered: The intended workload flows reach runtime controls while unauthorized lateral calls fail.
B:This is a valid design/control/evidence statement in the specified case, not the assumption to reject. Design a supported Prisma AIRS runtime placement and cluster segmentation to cover actual pod-to-model flows. It is consistent with the required outcome: The intended workload flows reach runtime controls while unauthorized lateral calls fail. Case R152-C028; angle 3; post-pilot control gap; option considered: Design a supported Prisma AIRS runtime placement and cluster segmentation to cover actual pod-to-model flows.
D:This is a valid design/control/evidence statement in the specified case, not the assumption to reject. Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path. It is consistent with the required outcome: The intended workload flows reach runtime controls while unauthorized lateral calls fail. Case R152-C028; angle 3; post-pilot control gap; option considered: Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path.
Question 112
Select TWO directly linked controls for North America regional bank: the architectural remedy and its decisive validation. After North America regional bank completed a limited security transformation pilot, reviewers observed a control gap: AI inference pods in Kubernetes share a trust domain and sensitive model-service calls bypass inspection. The case assumes the affected flows and dependencies have been traced and all unrelated systems are functioning; choose the narrow design repair before wider deployment.
- For North America regional bank: Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path.
- For North America regional bank: Add an internet edge firewall that never observes the intra-cluster model traffic.
- For North America regional bank: Choose Prisma AIRS runtime network-intercept architecture with an inline inspection/enforcement path and defined failure behavior.
- For North America regional bank: Design a supported Prisma AIRS runtime placement and cluster segmentation to cover actual pod-to-model flows.
- For North America regional bank: Send a synthetic confidential document through a test retrieval-to-model workflow and examine control verdicts.
Explanation
Correct Answer
A:This checks the relevant ai kubernetes east-west security enforcement path rather than a distant symptom. Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path. The expected evidence is The intended workload flows reach runtime controls while unauthorized lateral calls fail. Case R152-C028; angle 4; post-pilot control gap; option considered: Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path.
D:This directly addresses the recorded gap: AI inference pods in Kubernetes share a trust domain and sensitive model-service calls bypass inspection. The architectural control is Design a supported Prisma AIRS runtime placement and cluster segmentation to cover actual pod-to-model flows. The decisive test is Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path. The objective is The intended workload flows reach runtime controls while unauthorized lateral calls fail. Case R152-C028; angle 4; post-pilot control gap; option considered: Design a supported Prisma AIRS runtime placement and cluster segmentation to cover actual pod-to-model flows.
Incorrect Answers
B:This is an unsafe or insufficient shortcut for ai kubernetes east-west security: Add an internet edge firewall that never observes the intra-cluster model traffic. The scoped requirement instead calls for Design a supported Prisma AIRS runtime placement and cluster segmentation to cover actual pod-to-model flows. Case R152-C028; angle 4; post-pilot control gap; option considered: Add an internet edge firewall that never observes the intra-cluster model traffic.
C:This option addresses a different design objective: AI runtime network inspection, not the primary control for ai kubernetes east-west security. It would be relevant where An AI model endpoint must be protected without modifying application code, and all inference traffic crosses a controllable network path. Here the necessary evidence is Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path. Case R152-C028; angle 4; post-pilot control gap; option considered: Choose Prisma AIRS runtime network-intercept architecture with an inline inspection/enforcement path and defined failure behavior.
E:This option addresses a different design objective: In-flight AI data privacy, not the primary control for ai kubernetes east-west security. It would be relevant where A retrieval-augmented chatbot sends private documents in prompt context and compliance forbids unfiltered outbound data. Here the necessary evidence is Trace an authorized namespace-to-model request and a forbidden namespace request across the proposed inspection path. Case R152-C028; angle 4; post-pilot control gap; option considered: Send a synthetic confidential document through a test retrieval-to-model workflow and examine control verdicts.
Question 113
For Asia-Pacific SaaS provider, which is the most directly supported architectural control for shadow ai application data control? The security architecture board of Asia-Pacific SaaS provider is preparing a new AI Security program. The stakeholders have given the following specific requirement: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. No existing change is being reverted; select the most directly justified architecture for this stated requirement.
- For Asia-Pacific SaaS provider: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
- For Asia-Pacific SaaS provider: Choose Prisma AIRS runtime network-intercept architecture with an inline inspection/enforcement path and defined failure behavior.
- For Asia-Pacific SaaS provider: Integrate Prisma AIRS API Intercept through the supported API or SDK at the agent request/response path.
- For Asia-Pacific SaaS provider: Design a supported Prisma AIRS runtime placement and cluster segmentation to cover actual pod-to-model flows.
Explanation
Correct Answer
A:This directly addresses the recorded gap: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. The architectural control is Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. The decisive test is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. The objective is AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C029; angle 1; architecture design; option considered: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
Incorrect Answers
B:This option addresses a different design objective: AI runtime network inspection, not the primary control for shadow ai application data control. It would be relevant where An AI model endpoint must be protected without modifying application code, and all inference traffic crosses a controllable network path. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C029; angle 1; architecture design; option considered: Choose Prisma AIRS runtime network-intercept architecture with an inline inspection/enforcement path and defined failure behavior.
C:AI runtime API intercept addresses a different architectural concern, such as when a serverless AI agent needs prompt and response scanning but cannot be routed through a separate firewall hop. It does not establish shadow ai application data control here. The decisive test is to test an approved AI destination and a nonapproved destination using synthetic sensitive data.
D:This option addresses a different design objective: AI Kubernetes east-west security, not the primary control for shadow ai application data control. It would be relevant where AI inference pods in Kubernetes share a trust domain and sensitive model-service calls bypass inspection. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C029; angle 1; architecture design; option considered: Design a supported Prisma AIRS runtime placement and cluster segmentation to cover actual pod-to-model flows.
Question 114
Which validation most directly demonstrates shadow ai application data control is correctly designed for Asia-Pacific SaaS provider? The security architecture board of Asia-Pacific SaaS provider is preparing a new AI Security program. The stakeholders have given the following specific requirement: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. No existing change is being reverted; select the most directly justified architecture for this stated requirement.
- For Asia-Pacific SaaS provider: Exercise an authorized prompt-injection test suite and compare findings before and after a targeted fix.
- For Asia-Pacific SaaS provider: Submit a test model artifact with a known scan finding and inspect the predeployment decision record.
- For Asia-Pacific SaaS provider: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
- For Asia-Pacific SaaS provider: Send controlled safe and malicious model exchanges through the proposed network intercept and review enforcement telemetry.
Explanation
Correct Answer
C:This checks the relevant shadow ai application data control enforcement path rather than a distant symptom. Test an approved AI destination and a nonapproved destination using synthetic sensitive data. The expected evidence is AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C029; angle 2; architecture design; option considered: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
Incorrect Answers
A:AI red-team assessment addresses a different architectural concern, such as when a customer chatbot passed functional tests but its prompt-injection and jailbreak weaknesses were never evaluated. It does not establish shadow ai application data control here. The decisive test is to test an approved AI destination and a nonapproved destination using synthetic sensitive data.
B:This option addresses a different design objective: AI model supply-chain scanning, not the primary control for shadow ai application data control. It would be relevant where A model artifact from an external registry is deployed into production without provenance or risk evaluation. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C029; angle 2; architecture design; option considered: Submit a test model artifact with a known scan finding and inspect the predeployment decision record.
D:This option addresses a different design objective: AI runtime network inspection, not the primary control for shadow ai application data control. It would be relevant where An AI model endpoint must be protected without modifying application code, and all inference traffic crosses a controllable network path. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C029; angle 2; architecture design; option considered: Send controlled safe and malicious model exchanges through the proposed network intercept and review enforcement telemetry.
Question 115
Which design assumption should the architect REJECT when evaluating shadow ai application data control for Asia-Pacific SaaS provider? The security architecture board of Asia-Pacific SaaS provider is preparing a new AI Security program. The stakeholders have given the following specific requirement: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. No existing change is being reverted; select the most directly justified architecture for this stated requirement.
- For Asia-Pacific SaaS provider: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
- For Asia-Pacific SaaS provider: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement.
- For Asia-Pacific SaaS provider: Depend on the AI application's public marketing label as sufficient data-loss control.
- For Asia-Pacific SaaS provider: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
Explanation
Correct Answer
C:Reject this assumption: Depend on the AI application's public marketing label as sufficient data-loss control. It does not establish shadow ai application data control; the supported architecture is Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. Verify it by: Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C029; angle 3; architecture design; option considered: Depend on the AI application's public marketing label as sufficient data-loss control.
Incorrect Answers
A:This is a valid design/control/evidence statement in the specified case, not the assumption to reject. Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. It is consistent with the required outcome: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C029; angle 3; architecture design; option considered: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
B:This is a valid design/control/evidence statement in the specified case, not the assumption to reject. AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. It is consistent with the required outcome: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C029; angle 3; architecture design; option considered: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement.
D:This is a valid design/control/evidence statement in the specified case, not the assumption to reject. Test an approved AI destination and a nonapproved destination using synthetic sensitive data. It is consistent with the required outcome: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C029; angle 3; architecture design; option considered: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
Question 116
Select THREE linked elements for Asia-Pacific SaaS provider: a supported architecture, the direct test, and the required outcome. The security architecture board of Asia-Pacific SaaS provider is preparing a new AI Security program. The stakeholders have given the following specific requirement: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. No existing change is being reverted; select the most directly justified architecture for this stated requirement.
- For Asia-Pacific SaaS provider: Define agent identity, tool permissions, action scope, and AI runtime checks before privileged tool execution.
- For Asia-Pacific SaaS provider: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
- For Asia-Pacific SaaS provider: The intended workload flows reach runtime controls while unauthorized lateral calls fail.
- For Asia-Pacific SaaS provider: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement.
- For Asia-Pacific SaaS provider: Depend on the AI application's public marketing label as sufficient data-loss control.
- For Asia-Pacific SaaS provider: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
Explanation
Correct Answer
B:This checks the relevant shadow ai application data control enforcement path rather than a distant symptom. Test an approved AI destination and a nonapproved destination using synthetic sensitive data. The expected evidence is AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C029; angle 4; architecture design; option considered: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
D:The required end state is AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. The architect must deploy Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. and validate using Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C029; angle 4; architecture design; option considered: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement.
F:This directly addresses the recorded gap: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. The architectural control is Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. The decisive test is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. The objective is AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C029; angle 4; architecture design; option considered: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
Incorrect Answers
A:This option addresses a different design objective: Agent tool-call governance, not the primary control for shadow ai application data control. It would be relevant where A tool-using AI agent can act on privileged APIs even when user identity and task authorization are unverified. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C029; angle 4; architecture design; option considered: Define agent identity, tool permissions, action scope, and AI runtime checks before privileged tool execution.
C:This option addresses a different design objective: AI Kubernetes east-west security, not the primary control for shadow ai application data control. It would be relevant where AI inference pods in Kubernetes share a trust domain and sensitive model-service calls bypass inspection. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C029; angle 4; architecture design; option considered: The intended workload flows reach runtime controls while unauthorized lateral calls fail.
E:This is an unsafe or insufficient shortcut for shadow ai application data control: Depend on the AI application's public marketing label as sufficient data-loss control. The scoped requirement instead calls for Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. Case R152-C029; angle 4; architecture design; option considered: Depend on the AI application's public marketing label as sufficient data-loss control.
Question 117
For U.S. regional insurer, which is the most directly supported architectural control for shadow ai application data control? After U.S. regional insurer completed a limited security transformation pilot, reviewers observed a control gap: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. The case assumes the affected flows and dependencies have been traced and all unrelated systems are functioning; choose the narrow design repair before wider deployment.
- For U.S. regional insurer: Define agent identity, tool permissions, action scope, and AI runtime checks before privileged tool execution.
- For U.S. regional insurer: Map controls separately to AI Access and an appropriate Prisma AIRS runtime deployment rather than forcing one unsupported placement.
- For U.S. regional insurer: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
- For U.S. regional insurer: Choose Prisma AIRS runtime network-intercept architecture with an inline inspection/enforcement path and defined failure behavior.
Explanation
Correct Answer
C:This directly addresses the recorded gap: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. The architectural control is Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. The decisive test is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. The objective is AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C030; angle 1; post-pilot control gap; option considered: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
Incorrect Answers
A:This option addresses a different design objective: Agent tool-call governance, not the primary control for shadow ai application data control. It would be relevant where A tool-using AI agent can act on privileged APIs even when user identity and task authorization are unverified. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C030; angle 1; post-pilot control gap; option considered: Define agent identity, tool permissions, action scope, and AI runtime checks before privileged tool execution.
B:AI security deployment selection addresses a different architectural concern, such as when an architect must protect both browser-based enterprise AI use and a custom private model API with differing enforcement paths. It does not establish shadow ai application data control here. The decisive test is to test an approved AI destination and a nonapproved destination using synthetic sensitive data.
D:This option addresses a different design objective: AI runtime network inspection, not the primary control for shadow ai application data control. It would be relevant where An AI model endpoint must be protected without modifying application code, and all inference traffic crosses a controllable network path. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C030; angle 1; post-pilot control gap; option considered: Choose Prisma AIRS runtime network-intercept architecture with an inline inspection/enforcement path and defined failure behavior.
Question 118
Which validation most directly demonstrates shadow ai application data control is correctly designed for U.S. regional insurer? After U.S. regional insurer completed a limited security transformation pilot, reviewers observed a control gap: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. The case assumes the affected flows and dependencies have been traced and all unrelated systems are functioning; choose the narrow design repair before wider deployment.
- For U.S. regional insurer: Attempt the same permitted business operation in an approved workspace and in a disallowed app.
- For U.S. regional insurer: Replay an authorized low-risk tool request and a forged high-privilege request with distinct identity contexts.
- For U.S. regional insurer: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
- For U.S. regional insurer: Submit a test model artifact with a known scan finding and inspect the predeployment decision record.
Explanation
Correct Answer
C:This checks the relevant shadow ai application data control enforcement path rather than a distant symptom. Test an approved AI destination and a nonapproved destination using synthetic sensitive data. The expected evidence is AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C030; angle 2; post-pilot control gap; option considered: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
Incorrect Answers
A:Sanctioned AI app policy addresses a different architectural concern, such as when two departments use different AI apps but security cannot distinguish sanctioned workspaces from unknown services. It does not establish shadow ai application data control here. The decisive test is to test an approved AI destination and a nonapproved destination using synthetic sensitive data.
B:This option addresses a different design objective: Agent tool-call governance, not the primary control for shadow ai application data control. It would be relevant where A tool-using AI agent can act on privileged APIs even when user identity and task authorization are unverified. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C030; angle 2; post-pilot control gap; option considered: Replay an authorized low-risk tool request and a forged high-privilege request with distinct identity contexts.
D:This option addresses a different design objective: AI model supply-chain scanning, not the primary control for shadow ai application data control. It would be relevant where A model artifact from an external registry is deployed into production without provenance or risk evaluation. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C030; angle 2; post-pilot control gap; option considered: Submit a test model artifact with a known scan finding and inspect the predeployment decision record.
Question 119
Which design assumption should the architect REJECT when evaluating shadow ai application data control for U.S. regional insurer? After U.S. regional insurer completed a limited security transformation pilot, reviewers observed a control gap: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. The case assumes the affected flows and dependencies have been traced and all unrelated systems are functioning; choose the narrow design repair before wider deployment.
- For U.S. regional insurer: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
- For U.S. regional insurer: Depend on the AI application's public marketing label as sufficient data-loss control.
- For U.S. regional insurer: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
- For U.S. regional insurer: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement.
Explanation
Correct Answer
B:Reject this assumption: Depend on the AI application's public marketing label as sufficient data-loss control. It does not establish shadow ai application data control; the supported architecture is Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. Verify it by: Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C030; angle 3; post-pilot control gap; option considered: Depend on the AI application's public marketing label as sufficient data-loss control.
Incorrect Answers
A:This is a valid design/control/evidence statement in the specified case, not the assumption to reject. Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. It is consistent with the required outcome: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C030; angle 3; post-pilot control gap; option considered: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
C:This is a valid design/control/evidence statement in the specified case, not the assumption to reject. Test an approved AI destination and a nonapproved destination using synthetic sensitive data. It is consistent with the required outcome: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C030; angle 3; post-pilot control gap; option considered: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
D:This is a valid design/control/evidence statement in the specified case, not the assumption to reject. AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. It is consistent with the required outcome: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C030; angle 3; post-pilot control gap; option considered: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement.
Question 120
Select THREE linked elements for U.S. regional insurer: a supported architecture, the direct test, and the required outcome. After U.S. regional insurer completed a limited security transformation pilot, reviewers observed a control gap: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. The case assumes the affected flows and dependencies have been traced and all unrelated systems are functioning; choose the narrow design repair before wider deployment.
- For U.S. regional insurer: Design AI runtime data checks plus governed retrieval context and appropriate DLP enforcement for the chosen integration.
- For U.S. regional insurer: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement.
- For U.S. regional insurer: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
- For U.S. regional insurer: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
- For U.S. regional insurer: An unsafe or unapproved model is stopped before promotion and the finding remains auditable.
- For U.S. regional insurer: Depend on the AI application's public marketing label as sufficient data-loss control.
Explanation
Correct Answer
B:The required end state is AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. The architect must deploy Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. and validate using Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C030; angle 4; post-pilot control gap; option considered: AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement.
C:This checks the relevant shadow ai application data control enforcement path rather than a distant symptom. Test an approved AI destination and a nonapproved destination using synthetic sensitive data. The expected evidence is AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C030; angle 4; post-pilot control gap; option considered: Test an approved AI destination and a nonapproved destination using synthetic sensitive data.
D:This directly addresses the recorded gap: Employees paste confidential client data into unapproved AI websites and existing URL rules permit the traffic. The architectural control is Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. The decisive test is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. The objective is AI application use follows sanctioning decisions and sensitive uploads receive documented policy enforcement. Case R152-C030; angle 4; post-pilot control gap; option considered: Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels.
Incorrect Answers
A:In-flight AI data privacy addresses a different architectural concern, such as when a retrieval-augmented chatbot sends private documents in prompt context and compliance forbids unfiltered outbound data. It does not establish shadow ai application data control here. The decisive test is to test an approved AI destination and a nonapproved destination using synthetic sensitive data.
E:This option addresses a different design objective: AI model supply-chain scanning, not the primary control for shadow ai application data control. It would be relevant where A model artifact from an external registry is deployed into production without provenance or risk evaluation. Here the necessary evidence is Test an approved AI destination and a nonapproved destination using synthetic sensitive data. Case R152-C030; angle 4; post-pilot control gap; option considered: An unsafe or unapproved model is stopped before promotion and the finding remains auditable.
F:This is an unsafe or insufficient shortcut for shadow ai application data control: Depend on the AI application's public marketing label as sufficient data-loss control. The scoped requirement instead calls for Use AI app visibility and sanctioning with suitable application controls and Enterprise DLP inspection on supported channels. Case R152-C030; angle 4; post-pilot control gap; option considered: Depend on the AI application's public marketing label as sufficient data-loss control.
Palo Alto Networks Network Security Architect (NetSec-Architect) Practice Tests