McAfee-Secured Website

Exam Code: SnowPro Specialty Gen AI GES-C01

Exam Name: SnowPro Specialty Gen AI GES-C01

Certification Provider: Snowflake

100% Snowflake SnowPro Specialty Gen AI GES-C01 Guaranteed Success With Testking Training Materials

Get Real SnowPro Specialty Gen AI GES-C01 Exam Questions & Expert Verified Answers as Experienced on the Real Exam!

142 Questions & Answers with Testing Engine

"SnowPro Specialty Gen AI GES-C01 Exam", also known as SnowPro Specialty Gen AI GES-C01 exam, is a Snowflake certification exam.

Pass your tests with the always up-to-date SnowPro Specialty Gen AI GES-C01 Exam Engine. Your SnowPro Specialty Gen AI GES-C01 training materials keep you at the head of the pack!

Satisfaction Guaranteed

Money Back Guarantee

Testking provides hassle-free money back guarantee with our products. That is because we have 100% trust in the abilities of our professional and experience product team, and our record is a proof of that.

99.6% PASS RATE
Was: $137.49
Now: $124.99

Snowflake SnowPro Specialty Gen AI GES-C01 Practice Test 8

Question 70

At fictional Brookstream Commerce Alliance, a Snowflake data and AI engineering team is conducting a restricted enterprise language-model deployment under an approved model lifecycle and release-control process. The separately composed technical incident is: A private inference service runs in a container-backed workload, but the deployment plan exposes a development endpoint broadly and allows outbound calls unrelated to the model use case. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which repeatable assurance or control is most relevant to preventing this situation?

  1. For Brookstream Commerce Alliance during a restricted enterprise language-model deployment, version vector schemas, enforce model and dimension constraints, and rebenchmark retrieval before every embedding upgrade.
  2. For Brookstream Commerce Alliance during a restricted enterprise language-model deployment, add predeployment endpoint exposure scans, identity/egress reviews, resource quotas and periodic service-configuration drift checks.
  3. For Brookstream Commerce Alliance during a restricted enterprise language-model deployment, periodically rebalance the approved model matrix using actual production metrics, release changes and documented business acceptance tests.
  4. For Brookstream Commerce Alliance during a restricted enterprise language-model deployment, require environment-specific service identities, secret rotation, synthetic test data and a documented preproduction access-control validation.

Explanation

Writer-proposed answer: B (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

B:

Add predeployment endpoint exposure scans, identity/egress reviews, resource quotas and periodic service-configuration drift checks. Container-based model hosting still requires workload identity, network policy, secret hygiene and controlled service lifecycle.

Incorrect Answers

A:

This response concerns Embedding model migration consistency, which is different from the stated failure. A private inference service runs in a container-backed workload, but the deployment plan exposes a development endpoint broadly and allows outbound calls unrelated to the model use case.

C:

Model-selection latency and reliability tradeoffs is not the cause of the reported failure: A private inference service runs in a container-backed workload, but the deployment plan exposes a development endpoint broadly and allows outbound calls unrelated to the model use case.

D:

Separate development and production model rollout alone does not establish the needed evidence: Independent access tests reject anonymous and unauthorized endpoint calls, while service logs show only documented network destinations and safe secret handling.

Question 71

At fictional Greyson Customer Network, a Snowflake data and AI engineering team is conducting a regulated customer response automation exercise with regional service policies and accountable stakeholders. The separately composed technical incident is: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which authorized next action most directly addresses the stated system problem?

  1. For Greyson Customer Network during a regulated customer response automation exercise, record the embedding model and dimensions for each vector, rebuild or isolate indexes when representations change, and validate retrieval against a labeled evaluation set.
  2. For Greyson Customer Network during a regulated customer response automation exercise, register versioned model artifacts with provenance and evaluation metadata, restrict promotion rights and pin the approved version used by each service endpoint.
  3. For Greyson Customer Network during a regulated customer response automation exercise, treat retrieved text as untrusted evidence, separate it from higher-priority application instructions, refuse credential requests and require authorization at every tool call.
  4. For Greyson Customer Network during a regulated customer response automation exercise, design retrieval over authorized policy records, require answer citations to specific retrieved passages, and return a reviewable uncertainty response when no relevant source is found.

Explanation

Writer-proposed answer: C (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

C:

Retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary. Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Incorrect Answers

A:

This response concerns Embedding model migration consistency, which is different from the stated failure. A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

B:

Custom model artifact governance is insufficient to demonstrate this result: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

D:

Grounded employee policy assistant architecture alone does not establish the needed evidence: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Question 72

At fictional Greyson Customer Network, a Snowflake data and AI engineering team is conducting a regulated customer response automation exercise with regional service policies and accountable stakeholders. The separately composed technical incident is: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which independent evidence best establishes that the proposed corrective action worked?

  1. For Greyson Customer Network during a regulated customer response automation exercise, independent access tests reject anonymous and unauthorized endpoint calls, while service logs show only documented network destinations and safe secret handling.
  2. For Greyson Customer Network during a regulated customer response automation exercise, security tests show no server secrets in the delivered client assets and confirm unauthorized API calls cannot execute or access protected datasets.
  3. For Greyson Customer Network during a regulated customer response automation exercise, cross-role integration tests show that permitted questions succeed while requests for restricted records fail before retrieval or tool execution, with auditable denial evidence.
  4. For Greyson Customer Network during a regulated customer response automation exercise, adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Explanation

Writer-proposed answer: D (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

D:

Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials. Retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary.

Incorrect Answers

A:

This response concerns Snowpark Container Services endpoint isolation, which is different from the stated failure. A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

B:

SQL and REST authentication boundary is insufficient to demonstrate this result: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

C:

Agent permissions and external tool boundaries is not the cause of the reported failure: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

Question 73

At fictional Greyson Customer Network, a Snowflake data and AI engineering team is conducting a regulated customer response automation exercise with regional service policies and accountable stakeholders. The separately composed technical incident is: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which architecture or data-governance principle best explains the scenario and solution?

  1. For Greyson Customer Network during a regulated customer response automation exercise, retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary.
  2. For Greyson Customer Network during a regulated customer response automation exercise, model registry governance provides an accountable relationship between training artifacts, evaluated versions and serving decisions.
  3. For Greyson Customer Network during a regulated customer response automation exercise, environment isolation and controlled promotion reduce the blast radius of Gen AI experimentation without sacrificing testability.
  4. For Greyson Customer Network during a regulated customer response automation exercise, model suitability depends on the task, governance, performance and cost rather than parameter size or marketing claims alone.

Explanation

Writer-proposed answer: A (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

A:

Retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary. Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Incorrect Answers

B:

Custom model artifact governance is insufficient to demonstrate this result: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

C:

Separate development and production model rollout is not the cause of the reported failure: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

D:

Model-selection latency and reliability tradeoffs alone does not establish the needed evidence: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Question 74

At fictional Greyson Customer Network, a Snowflake data and AI engineering team is conducting a regulated customer response automation exercise with regional service policies and accountable stakeholders. The separately composed technical incident is: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which untreated operational or security risk most specifically follows from this scenario?

  1. For Greyson Customer Network during a regulated customer response automation exercise, unreviewed cross-region fallback could transmit protected information across a prohibited jurisdiction or create untested resilience assumptions.
  2. For Greyson Customer Network during a regulated customer response automation exercise, following embedded instructions can hijack tool selection, disclose secrets or silently corrupt an answer while appearing grounded.
  3. For Greyson Customer Network during a regulated customer response automation exercise, an overfit demo can appear reliable while failing on rare cases, missing facts and adversarial prompts in production.
  4. For Greyson Customer Network during a regulated customer response automation exercise, without grounded retrieval and an abstention path, fabricated policy statements could become operational decisions with no accountable source.

Explanation

Writer-proposed answer: B (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

B:

Following embedded instructions can hijack tool selection, disclose secrets or silently corrupt an answer while appearing grounded. Retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary.

Incorrect Answers

A:

This response concerns Cross-region model inference consent and residency, which is different from the stated failure. A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

C:

Gen AI evaluation set representativeness is not the cause of the reported failure: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

D:

Grounded employee policy assistant architecture alone does not establish the needed evidence: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Question 75

At fictional Greyson Customer Network, a Snowflake data and AI engineering team is conducting a regulated customer response automation exercise with regional service policies and accountable stakeholders. The separately composed technical incident is: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which repeatable assurance or control is most relevant to preventing this situation?

  1. For Greyson Customer Network during a regulated customer response automation exercise, require environment-specific service identities, secret rotation, synthetic test data and a documented preproduction access-control validation.
  2. For Greyson Customer Network during a regulated customer response automation exercise, record supported intent types, test routing decisions against signed business metrics and flag unanswered questions for semantic-model or retrieval-coverage improvements.
  3. For Greyson Customer Network during a regulated customer response automation exercise, include malicious-document tests, source-trust labels and tool-layer authorization checks in each retrieval pipeline release.
  4. For Greyson Customer Network during a regulated customer response automation exercise, require realistic load tests, backpressure and retry budgets, alerting on tail latency and service usage, and periodic capacity review.

Explanation

Writer-proposed answer: C (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

C:

Include malicious-document tests, source-trust labels and tool-layer authorization checks in each retrieval pipeline release. Retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary.

Incorrect Answers

A:

This response concerns Separate development and production model rollout, which is different from the stated failure. A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

B:

Structured analytics versus document search routing is insufficient to demonstrate this result: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

D:

LLM and retrieval service capacity budgeting alone does not establish the needed evidence: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Question 76

At fictional Newcastle Infrastructure Lab, a Snowflake data and AI engineering team is conducting a regional AI data processing evaluation using only first-party public technical references and independently written facts. The separately composed technical incident is: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which authorized next action most directly addresses the stated system problem?

  1. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, bind each agent tool to the minimum approved privileges, preserve the requesting user's authorization context, and deny cross-domain access before invoking analytical or search services.
  2. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, define a minimally exposed service endpoint, constrain runtime roles, network egress and secrets, and test authentication and resource isolation before approving access.
  3. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, design retrieval over authorized policy records, require answer citations to specific retrieved passages, and return a reviewable uncertainty response when no relevant source is found.
  4. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, treat retrieved text as untrusted evidence, separate it from higher-priority application instructions, refuse credential requests and require authorization at every tool call.

Explanation

Writer-proposed answer: D (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

D:

Retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary. Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Incorrect Answers

A:

This response concerns Agent permissions and external tool boundaries, which is different from the stated failure. A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

B:

Snowpark Container Services endpoint isolation is insufficient to demonstrate this result: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

C:

Grounded employee policy assistant architecture is not the cause of the reported failure: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

Question 77

At fictional Newcastle Infrastructure Lab, a Snowflake data and AI engineering team is conducting a regional AI data processing evaluation using only first-party public technical references and independently written facts. The separately composed technical incident is: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which independent evidence best establishes that the proposed corrective action worked?

  1. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.
  2. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, a curated test set of aggregate and narrative questions returns correct reconciled warehouse metrics from structured queries and source-linked passages from documents.
  3. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, concurrent load tests meet approved tail-latency and error objectives, and observed credit consumption reconciles to the approved workload budget.
  4. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, independent test prompts with present, absent and conflicting policies show cited supporting passages for grounded answers and explicit abstention for unsupported claims.

Explanation

Writer-proposed answer: A (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

A:

Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials. Retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary.

Incorrect Answers

B:

Structured analytics versus document search routing is insufficient to demonstrate this result: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

C:

LLM and retrieval service capacity budgeting is not the cause of the reported failure: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

D:

Grounded employee policy assistant architecture alone does not establish the needed evidence: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Question 78

At fictional Newcastle Infrastructure Lab, a Snowflake data and AI engineering team is conducting a regional AI data processing evaluation using only first-party public technical references and independently written facts. The separately composed technical incident is: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which architecture or data-governance principle best explains the scenario and solution?

  1. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, a language model's fluency does not establish factual grounding; retrieval, source attribution and evidence-based refusal must be evaluated separately.
  2. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary.
  3. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, model selection and regional availability must be evaluated together with data residency, latency, governance and legal obligations.
  4. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, a representative evaluation must detect generalization failures, not reward memorization of tuning examples.

Explanation

Writer-proposed answer: B (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

B:

Retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary. Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Incorrect Answers

A:

This response concerns Grounded employee policy assistant architecture, which is different from the stated failure. A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

C:

Cross-region model inference consent and residency is not the cause of the reported failure: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

D:

Gen AI evaluation set representativeness alone does not establish the needed evidence: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Question 79

At fictional Newcastle Infrastructure Lab, a Snowflake data and AI engineering team is conducting a regional AI data processing evaluation using only first-party public technical references and independently written facts. The separately composed technical incident is: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which untreated operational or security risk most specifically follows from this scenario?

  1. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, unbounded parallel requests and retry storms can raise cost, degrade availability and make critical workflows unavailable.
  2. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, oversized or unsuitable models can cause unpredictable costs, timeouts, failed JSON integrations and regional policy violations.
  3. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, following embedded instructions can hijack tool selection, disclose secrets or silently corrupt an answer while appearing grounded.
  4. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, shared identities and secrets can disclose production records and allow unreviewed prompts or model settings to reach customers.

Explanation

Writer-proposed answer: C (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

C:

Following embedded instructions can hijack tool selection, disclose secrets or silently corrupt an answer while appearing grounded. Retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary.

Incorrect Answers

A:

This response concerns LLM and retrieval service capacity budgeting, which is different from the stated failure. A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

B:

Model-selection latency and reliability tradeoffs is insufficient to demonstrate this result: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

D:

Separate development and production model rollout alone does not establish the needed evidence: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

Question 80

At fictional Newcastle Infrastructure Lab, a Snowflake data and AI engineering team is conducting a regional AI data processing evaluation using only first-party public technical references and independently written facts. The separately composed technical incident is: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which repeatable assurance or control is most relevant to preventing this situation?

  1. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, automate client-secret scanning, scoped credential issuance, role reviews and emergency rotation drills for integration identities.
  2. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, add per-tool role reviews, least-privilege integration tests and approval gates for any new tool, search index or stored procedure connected to the agent.
  3. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, maintain a regression set of missing-source and conflicting-source queries, review citation coverage and update retrieval permissions whenever the policy repository changes.
  4. For Newcastle Infrastructure Lab during a regional AI data processing evaluation, include malicious-document tests, source-trust labels and tool-layer authorization checks in each retrieval pipeline release.

Explanation

Writer-proposed answer: D (UNVERIFIED)

Preliminary individual option explanations — NOT expert validated

Correct Answer

D:

Include malicious-document tests, source-trust labels and tool-layer authorization checks in each retrieval pipeline release. Retrieved documents are lower-trust content, not authenticated instructions; safe retrieval-augmented generation maintains a clear instruction/data boundary.

Incorrect Answers

A:

This response concerns SQL and REST authentication boundary, which is different from the stated failure. A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

B:

Agent permissions and external tool boundaries is insufficient to demonstrate this result: Adversarial document-injection tests show that the agent cites the document as data but never follows its malicious commands or fetches secret credentials.

C:

Grounded employee policy assistant architecture is not the cause of the reported failure: A support assistant retrieves a vendor document that contains instructions to ignore access controls and disclose internal credentials; the application currently includes document text in the same instruction channel as its own rules.

SnowPro Specialty Gen AI GES-C01 Practice Tests