Cisco 810-110 Practice Test 18
Question 170
An adversary uses search-result text to induce an unrelated outbound network action. Which practical correction or implementation decision is MOST directly appropriate?
- Treat an address found in untrusted text as explicit user approval to email secrets.
- Isolate retrieved content and constrain side-effect tools to user-authorized tasks.
- Maintain trusted instruction hierarchy and harden tool authorization.
- Enforce least-privilege tools and user authorization for all outbound sharing.
Explanation
Correct Answer
D:Enforce least-privilege tools and user authorization for all outbound sharing. The intended behavior can be checked because no protected content is sent to an unapproved recipient in attack simulations.
Incorrect Answers
A:Treat an address found in untrusted text as explicit user approval to email secrets. This would not correct the Injection-driven data exfiltration workflow. Trusting a document-defined destination can leak confidential data.
B:Isolate retrieved content and constrain side-effect tools to user-authorized tasks. This could help Indirect injection in retrieved webpages, but it does not resolve the Injection-driven data exfiltration case. Prompt injection can target tool permissions, not only the content of a model's answer.
C:Maintain trusted instruction hierarchy and harden tool authorization. This could help Direct user instruction override attempt, but it does not resolve the Injection-driven data exfiltration case. Prompt injection can target tool permissions, not only the content of a model's answer.
Question 171
A model-training table has blank incomes and unknown city values. After the recommended Missing-value strategies and leakage prevention change is implemented, which observation MOST directly verifies that it works?
- A second analyst reproduces the same summary from the recorded inputs.
- Test data is transformed without using hidden future or holdout statistics.
- Quality alerts identify the affected batch before downstream reports are trusted.
- Parsed records meet the chosen schema with flagged invalid values.
Explanation
Correct Answer
B:Test data is transformed without using hidden future or holdout statistics. This directly tests whether the Missing-value strategies and leakage prevention goal was met. Imputation must reflect variable semantics and avoid training-test leakage.
Incorrect Answers
A:A second analyst reproduces the same summary from the recorded inputs. This is evidence about Reproducible EDA with provenance, not direct proof of the required Missing-value strategies and leakage prevention behavior. The relevant check is that test data is transformed without using hidden future or holdout statistics.
C:Quality alerts identify the affected batch before downstream reports are trusted. This is evidence about Automated data-quality drift checks, not direct proof of the required Missing-value strategies and leakage prevention behavior. The relevant check is that test data is transformed without using hidden future or holdout statistics.
D:Parsed records meet the chosen schema with flagged invalid values. This is evidence about Type-aware normalization of input data, not direct proof of the required Missing-value strategies and leakage prevention behavior. The relevant check is that test data is transformed without using hidden future or holdout statistics.
Question 172
An AI coding assistant must edit one parser in a large monorepo. Which approach BEST meets the requirement while respecting the system constraints?
- Remove every dependency interface from the prompt so the model can guess it.
- Preserve exact interfaces, constraints, and tests in a compact structured state.
- Index the code and retrieve only relevant modules, interfaces, and tests.
- Check finish reasons and parse completeness before accepting generated outputs.
Explanation
Correct Answer
C:Index the code and retrieve only relevant modules, interfaces, and tests. Selective context improves both signal-to-noise ratio and token use. Verification should demonstrate that generated edits reference actual APIs and fit in the available context.
Incorrect Answers
A:Remove every dependency interface from the prompt so the model can guess it. This is an unsafe or ineffective shortcut for Repository chunk selection. Including every unrelated file can obscure the target function or truncate tests.
B:Preserve exact interfaces, constraints, and tests in a compact structured state. This primarily addresses Compression and summary quality risks, not the specific Repository chunk selection decision. Here, selective context improves both signal-to-noise ratio and token use.
D:Check finish reasons and parse completeness before accepting generated outputs. This primarily addresses Detecting output truncation, not the specific Repository chunk selection decision. Here, selective context improves both signal-to-noise ratio and token use.
Question 173
A temporary research agent needs selected rows from a confidential finance database. Which approach BEST meets the requirement while respecting the system constraints?
- Apply query-rate, minimum-group, and disclosure controls to prevent differencing attacks.
- Apply minimum cohort size rules and suppress identifying small cells.
- Run exploratory generated code with administrator access against live finance tables.
- Supply a least-privilege sanitized sandbox dataset with controlled export.
Explanation
Correct Answer
D:Supply a least-privilege sanitized sandbox dataset with controlled export. Sandboxed data access limits blast radius during experimentation. Verification should demonstrate that the prototype works without direct broad production access or secret leakage.
Incorrect Answers
A:Apply query-rate, minimum-group, and disclosure controls to prevent differencing attacks. This primarily addresses Preventing inference through repeated queries, not the specific Secure analytics sandboxing decision. Here, sandboxed data access limits blast radius during experimentation.
B:Apply minimum cohort size rules and suppress identifying small cells. This primarily addresses Privacy-preserving aggregated reporting, not the specific Secure analytics sandboxing decision. Here, sandboxed data access limits blast radius during experimentation.
C:Run exploratory generated code with administrator access against live finance tables. This is an unsafe or ineffective shortcut for Secure analytics sandboxing. Giving unrestricted production credentials to a research prompt can leak or modify records.
Question 174
An enterprise wants traceability for who authorized an AI-driven configuration update. Which practical correction or implementation decision is MOST directly appropriate?
- Discard approval identity once the agent reports the action was done.
- Define confidence and risk thresholds for handoff with full context.
- Persist safe approval and completion receipts with scoped retention.
- Require verified explicit authorization before dangerous operations.
Explanation
Correct Answer
C:Persist safe approval and completion receipts with scoped retention. The intended behavior can be checked because an audit reviewer can reconstruct the authorized modification without raw secrets.
Incorrect Answers
A:Discard approval identity once the agent reports the action was done. This would not correct the Auditable checkpoint after consequential actions workflow. Keeping only a generated statement approved provides no reliable audit record.
B:Define confidence and risk thresholds for handoff with full context. This could help Low-confidence escalation to a specialist, but it does not resolve the Auditable checkpoint after consequential actions case. Human oversight needs evidence of both permission and actual execution.
D:Require verified explicit authorization before dangerous operations. This could help Fail-closed safety for destructive tools, but it does not resolve the Auditable checkpoint after consequential actions case. Human oversight needs evidence of both permission and actual execution.
Question 175
An assistant follows an instruction saying always cite sources but some sources are fictional. Which approach BEST meets the requirement while respecting the system constraints?
- Consider a citation valid simply because it contains a URL-shaped substring.
- Treat response instructions as formatting guidance and independently verify factual grounding.
- State evidence gaps and ask for missing context instead of fabricating specifics.
- Validate incoming text and generated structured outputs against domain rules before side effects.
Explanation
Correct Answer
B:Treat response instructions as formatting guidance and independently verify factual grounding. Prompt instructions cannot guarantee truth without evidence checking. Verification should demonstrate that cited passages actually exist and substantiate the statements they accompany.
Incorrect Answers
A:Consider a citation valid simply because it contains a URL-shaped substring. This is an unsafe or ineffective shortcut for Separation of prompt guidance from evidence verification. Automatically accepting source-shaped strings lets unsupported claims pass.
C:State evidence gaps and ask for missing context instead of fabricating specifics. This primarily addresses Appropriate uncertainty and abstention, not the specific Separation of prompt guidance from evidence verification decision. Here, prompt instructions cannot guarantee truth without evidence checking.
D:Validate incoming text and generated structured outputs against domain rules before side effects. This primarily addresses Input and output validation guardrails, not the specific Separation of prompt guidance from evidence verification decision. Here, prompt instructions cannot guarantee truth without evidence checking.
Question 176
A compliance officer asks why an automated assistant changed an account status. If the relevant Traceable AI decision audit logs safeguard is omitted, which outcome is the MOST directly associated failure?
- Without provenance and action logs, harmful automation is difficult to reconstruct.
- Installing invented package names can execute attacker-controlled code.
- Assuming the vendor handles every legal and security duty leaves gaps.
- Automatically publishing generated compliance claims can create false representations.
Explanation
Correct Answer
A:Without provenance and action logs, harmful automation is difficult to reconstruct. Audit trails support accountability while requiring their own access and privacy controls. The corrective approach is to maintain protected decision traces with retention and redaction.
Incorrect Answers
B:Installing invented package names can execute attacker-controlled code. This failure pattern belongs chiefly to AI-generated dependency security risks; it is not the failure most directly tested in the Traceable AI decision audit logs situation. Without provenance and action logs, harmful automation is difficult to reconstruct.
C:Assuming the vendor handles every legal and security duty leaves gaps. This failure pattern belongs chiefly to Vendor and model governance review; it is not the failure most directly tested in the Traceable AI decision audit logs situation. Without provenance and action logs, harmful automation is difficult to reconstruct.
D:Automatically publishing generated compliance claims can create false representations. This failure pattern belongs chiefly to Documented human approval for regulated changes; it is not the failure most directly tested in the Traceable AI decision audit logs situation. Without provenance and action logs, harmful automation is difficult to reconstruct.
Question 177
A security analyst uses a model summary to close an incident without reviewing original evidence. An engineer must explain the recommendation for Automation bias and overreliance. Which underlying principle is the MOST accurate?
- Untrusted content can redirect tool use unless trust boundaries are enforced.
- Fluent AI outputs can induce excessive trust even when they are incorrect.
- Corrupted training examples can steer learned behavior or embed backdoors.
- Excluding an explicit protected column does not eliminate proxy discrimination.
Explanation
Correct Answer
B:Fluent AI outputs can induce excessive trust even when they are incorrect. Therefore the implementation decision is to require corroborating evidence and accountable human judgment for material decisions.
Incorrect Answers
A:Uncritical acceptance can bury serious security or operational faults. Untrusted content can redirect tool use unless trust boundaries are enforced.
C:Uncritical acceptance can bury serious security or operational faults. Corrupted training examples can steer learned behavior or embed backdoors.
D:Uncritical acceptance can bury serious security or operational faults. Excluding an explicit protected column does not eliminate proxy discrimination.
Question 178
A model team ingests third-party training examples without source verification. Which practical correction or implementation decision is MOST directly appropriate?
- Resolve dependencies against verified registries and review AI code.
- Assume a dataset is safe because its CSV format parses successfully.
- Limit sensitive training content and test leakage with governed access.
- Use provenance checks, curation, and held-out adversarial validation.
Explanation
Correct Answer
D:Use provenance checks, curation, and held-out adversarial validation. The intended behavior can be checked because poisoned samples are detected or excluded before training and evaluation.
Incorrect Answers
A:Resolve dependencies against verified registries and review AI code. This could help AI-generated dependency security risks, but it does not resolve the Training data poisoning case. Corrupted training examples can steer learned behavior or embed backdoors.
B:Assume a dataset is safe because its CSV format parses successfully. This would not correct the Training data poisoning workflow. Blindly mixing attacker-controlled records can compromise model responses.
C:Limit sensitive training content and test leakage with governed access. This could help Sensitive data memorization exposure, but it does not resolve the Training data poisoning case. Corrupted training examples can steer learned behavior or embed backdoors.
Question 179
A tool returns a success status but omits an optional address field required by a later branch. If the relevant Robust handling of partial tool responses safeguard is omitted, which outcome is the MOST directly associated failure?
- Dropping event identity can allow duplicate writes after retries.
- Treating a local time string as UTC can schedule the wrong appointment.
- A guessed yes response can cause irreversible data loss.
- Blindly dereferencing missing keys or inventing values causes failures and incorrect records.
Explanation
Correct Answer
D:Blindly dereferencing missing keys or inventing values causes failures and incorrect records. Agents must treat external tool payloads as uncertain structured data. The corrective approach is to use schema validation and explicit missing-field recovery.
Incorrect Answers
A:Dropping event identity can allow duplicate writes after retries. This failure pattern belongs chiefly to Idempotent event identity across agents; it is not the failure most directly tested in the Robust handling of partial tool responses situation. Blindly dereferencing missing keys or inventing values causes failures and incorrect records.
B:Treating a local time string as UTC can schedule the wrong appointment. This failure pattern belongs chiefly to Date and timezone normalization; it is not the failure most directly tested in the Robust handling of partial tool responses situation. Blindly dereferencing missing keys or inventing values causes failures and incorrect records.
C:A guessed yes response can cause irreversible data loss. This failure pattern belongs chiefly to Fail-closed safety for destructive tools; it is not the failure most directly tested in the Robust handling of partial tool responses situation. Blindly dereferencing missing keys or inventing values causes failures and incorrect records.
Question 180
An enterprise wants traceability for who authorized an AI-driven configuration update. An engineer must explain the recommendation for Auditable checkpoint after consequential actions. Which underlying principle is the MOST accurate?
- Fail-closed policy prevents uncertainty from becoming unintended permission.
- Tool descriptions guide models but schemas and runtime checks enforce constraints.
- Human oversight needs evidence of both permission and actual execution.
- Human control requires meaningful intervention, not merely passive observation.
Explanation
Correct Answer
C:Human oversight needs evidence of both permission and actual execution. Therefore the implementation decision is to record action request, approver identity, decision, tool result, and verification evidence.
Incorrect Answers
A:Keeping only a generated statement approved provides no reliable audit record. This describes the separate Fail-closed safety for destructive tools mechanism; the present case instead requires Auditable checkpoint after consequential actions.
B:Keeping only a generated statement approved provides no reliable audit record. Tool descriptions guide models but schemas and runtime checks enforce constraints.
D:Keeping only a generated statement approved provides no reliable audit record. This describes the separate Human override and reversible control mechanism; the present case instead requires Auditable checkpoint after consequential actions.