ISACA Advanced in AI Audit AAIA Practice Test 20
Question 190
Sensitive fields are removed from a table, but a hashed identifier still links customers across public datasets. From an AI audit perspective, which corrective action is the MOST direct response to the weakness described, rather than a useful control for some different AI risk?
- Reassess linkage risk and protect keys, quasi-identifiers and dataset releases
- Introduce contract tests for unit, range and meaning changes before retraining
- Enforce typed identifiers and record-level reconciliation before feature merges
- Quantify segment coverage and apply justified sampling with separate evaluation
Explanation
Correct Answer
A:Correct. Reassess linkage risk and protect keys, quasi-identifiers and dataset releases is aimed at the actual failure: pseudonymization was treated as irreversibly anonymous. The audit should subsequently obtain reidentification tests, key access records and release-risk signoff to test execution, not merely the existence of a policy.
Incorrect Answers
B:“Introduce contract tests for unit, range and meaning changes before retraining” is a control for data-quality gates do not validate semantic feature definitions across pipeline versions, but the audit finding here is pseudonymization was treated as irreversibly anonymous. The relevant measure is “Reassess linkage risk and protect keys, quasi-identifiers and dataset releases”.
C:“Enforce typed identifiers and record-level reconciliation before feature merges” is a control for join normalization corrupts entity matching between data sources, but the audit finding here is pseudonymization was treated as irreversibly anonymous. The relevant measure is “Reassess linkage risk and protect keys, quasi-identifiers and dataset releases”.
D:“Quantify segment coverage and apply justified sampling with separate evaluation” is a control for representation imbalance hides subgroup error and instability, but the audit finding here is pseudonymization was treated as irreversibly anonymous. The relevant measure is “Reassess linkage risk and protect keys, quasi-identifiers and dataset releases”.
Question 191
In an audit of Data Management Specific to AI, managers say this measure is live: Reassess linkage risk and protect keys, quasi-identifiers and dataset releases. They claim it corrected a material audit gap. What direct operating evidence would allow the auditor to substantiate that particular claim across a defensible sample?
- Schema diff, unit conversion assertions and rejected malformed batches
- Join exception statistics and sampled identifier matches against source systems
- Per-segment counts, sampling decisions and stable held-out subgroup metrics
- Reidentification tests, key access records and release-risk signoff
Explanation
Correct Answer
D:Correct. The proposed records—reidentification tests, key access records and release-risk signoff—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is pseudonymization was treated as irreversibly anonymous.
Incorrect Answers
A:“Schema diff, unit conversion assertions and rejected malformed batches” would test data-quality gates do not validate semantic feature definitions across pipeline versions, not the present safeguard against pseudonymization was treated as irreversibly anonymous. The expected audit evidence is “Reidentification tests, key access records and release-risk signoff”.
B:“Join exception statistics and sampled identifier matches against source systems” would test join normalization corrupts entity matching between data sources, not the present safeguard against pseudonymization was treated as irreversibly anonymous. The expected audit evidence is “Reidentification tests, key access records and release-risk signoff”.
C:“Per-segment counts, sampling decisions and stable held-out subgroup metrics” would test representation imbalance hides subgroup error and instability, not the present safeguard against pseudonymization was treated as irreversibly anonymous. The expected audit evidence is “Reidentification tests, key access records and release-risk signoff”.
Question 192
In a review of Data Management Specific to AI, investigators cannot establish the expected control proof: reidentification tests, key access records and release-risk signoff. The underlying event was that sensitive fields are removed from a table, but a hashed identifier still links customers across public datasets. Which control weakness is the MOST likely cause requiring focused audit work?
- Data-quality gates do not validate semantic feature definitions across pipeline versions
- Join normalization corrupts entity matching between data sources
- Pseudonymization was treated as irreversibly anonymous
- Representation imbalance hides subgroup error and instability
Explanation
Correct Answer
C:Correct. The observed condition specifically indicates pseudonymization was treated as irreversibly anonymous. A meaningful audit response is to verify this diagnosis and then recommend reassess linkage risk and protect keys, quasi-identifiers and dataset releases.
Incorrect Answers
A:“Data-quality gates do not validate semantic feature definitions across pipeline versions” concerns a separate audit weakness, not the missing assurance described by pseudonymization was treated as irreversibly anonymous. The correct cause is “Pseudonymization was treated as irreversibly anonymous”.
B:“Join normalization corrupts entity matching between data sources” concerns a separate audit weakness, not the missing assurance described by pseudonymization was treated as irreversibly anonymous. The correct cause is “Pseudonymization was treated as irreversibly anonymous”.
D:“Representation imbalance hides subgroup error and instability” concerns a separate audit weakness, not the missing assurance described by pseudonymization was treated as irreversibly anonymous. The correct cause is “Pseudonymization was treated as irreversibly anonymous”.
Question 193
A training pipeline moves labeled records between regions without encryption or dataset-level access logs. From an AI audit perspective, which corrective action is the MOST direct response to the weakness described, rather than a useful control for some different AI risk?
- Audit label completeness by merchant age and use independently confirmed outcome samples
- Apply encrypted transport, integrity verification and attributable transfer logging
- Enforce event-time feature availability and cutoff validation for every predictor
- Review contractual source rights, permitted purpose and collection safeguards before use
Explanation
Correct Answer
B:Correct. Apply encrypted transport, integrity verification and attributable transfer logging is aimed at the actual failure: in-transit protection and accountable data movement are missing. The audit should subsequently obtain transfer telemetry, cryptographic configuration and tamper-detection evidence to test execution, not merely the existence of a policy.
Incorrect Answers
A:“Audit label completeness by merchant age and use independently confirmed outcome samples” is a control for missing labels and coverage create misleadingly low observed fraud rates, but the audit finding here is in-transit protection and accountable data movement are missing. The relevant measure is “Apply encrypted transport, integrity verification and attributable transfer logging”.
C:“Enforce event-time feature availability and cutoff validation for every predictor” is a control for outcome-period data leaks into predictors unavailable at scoring time, but the audit finding here is in-transit protection and accountable data movement are missing. The relevant measure is “Apply encrypted transport, integrity verification and attributable transfer logging”.
D:“Review contractual source rights, permitted purpose and collection safeguards before use” is a control for external data sourcing controls omit restrictions on collection and use, but the audit finding here is in-transit protection and accountable data movement are missing. The relevant measure is “Apply encrypted transport, integrity verification and attributable transfer logging”.
Question 194
In an audit of Data Management Specific to AI, managers say this measure is live: Apply encrypted transport, integrity verification and attributable transfer logging. They claim it corrected a material audit gap. What direct operating evidence would allow the auditor to substantiate that particular claim across a defensible sample?
- Transfer telemetry, cryptographic configuration and tamper-detection evidence
- Label reconciliation by tenure cohort and corrected training-data completeness report
- Point-in-time feature store queries compared with historical scoring timestamps
- Source inventory, collection logs, license terms and exception approval
Explanation
Correct Answer
A:Correct. The proposed records—transfer telemetry, cryptographic configuration and tamper-detection evidence—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is in-transit protection and accountable data movement are missing.
Incorrect Answers
B:“Label reconciliation by tenure cohort and corrected training-data completeness report” would test missing labels and coverage create misleadingly low observed fraud rates, not the present safeguard against in-transit protection and accountable data movement are missing. The expected audit evidence is “Transfer telemetry, cryptographic configuration and tamper-detection evidence”.
C:“Point-in-time feature store queries compared with historical scoring timestamps” would test outcome-period data leaks into predictors unavailable at scoring time, not the present safeguard against in-transit protection and accountable data movement are missing. The expected audit evidence is “Transfer telemetry, cryptographic configuration and tamper-detection evidence”.
D:“Source inventory, collection logs, license terms and exception approval” would test external data sourcing controls omit restrictions on collection and use, not the present safeguard against in-transit protection and accountable data movement are missing. The expected audit evidence is “Transfer telemetry, cryptographic configuration and tamper-detection evidence”.
Question 195
In a review of Data Management Specific to AI, investigators cannot establish the expected control proof: transfer telemetry, cryptographic configuration and tamper-detection evidence. The underlying event was that a training pipeline moves labeled records between regions without encryption or dataset-level access logs. Which control weakness is the MOST likely cause requiring focused audit work?
- Missing labels and coverage create misleadingly low observed fraud rates
- Outcome-period data leaks into predictors unavailable at scoring time
- External data sourcing controls omit restrictions on collection and use
- In-transit protection and accountable data movement are missing
Explanation
Correct Answer
D:Correct. The observed condition specifically indicates in-transit protection and accountable data movement are missing. A meaningful audit response is to verify this diagnosis and then recommend apply encrypted transport, integrity verification and attributable transfer logging.
Incorrect Answers
A:“Missing labels and coverage create misleadingly low observed fraud rates” concerns a separate audit weakness, not the missing assurance described by in-transit protection and accountable data movement are missing. The correct cause is “In-transit protection and accountable data movement are missing”.
B:“Outcome-period data leaks into predictors unavailable at scoring time” concerns a separate audit weakness, not the missing assurance described by in-transit protection and accountable data movement are missing. The correct cause is “In-transit protection and accountable data movement are missing”.
C:“External data sourcing controls omit restrictions on collection and use” concerns a separate audit weakness, not the missing assurance described by in-transit protection and accountable data movement are missing. The correct cause is “In-transit protection and accountable data movement are missing”.
Question 196
The data owner signs off on records before a schema change that silently converts currency units. From an AI audit perspective, which corrective action is the MOST direct response to the weakness described, rather than a useful control for some different AI risk?
- Encode missingness and validate imputation against sensor-state evidence
- Version immutable datasets and hashes linked to each model build
- Introduce contract tests for unit, range and meaning changes before retraining
- Reassess linkage risk and protect keys, quasi-identifiers and dataset releases
Explanation
Correct Answer
C:Correct. Introduce contract tests for unit, range and meaning changes before retraining is aimed at the actual failure: data-quality gates do not validate semantic feature definitions across pipeline versions. The audit should subsequently obtain schema diff, unit conversion assertions and rejected malformed batches to test execution, not merely the existence of a policy.
Incorrect Answers
A:“Encode missingness and validate imputation against sensor-state evidence” is a control for imputation logic confuses sensor outages with genuine zero measurements, but the audit finding here is data-quality gates do not validate semantic feature definitions across pipeline versions. The relevant measure is “Introduce contract tests for unit, range and meaning changes before retraining”.
B:“Version immutable datasets and hashes linked to each model build” is a control for data reproducibility and provenance are absent from the ML pipeline, but the audit finding here is data-quality gates do not validate semantic feature definitions across pipeline versions. The relevant measure is “Introduce contract tests for unit, range and meaning changes before retraining”.
D:“Reassess linkage risk and protect keys, quasi-identifiers and dataset releases” is a control for pseudonymization was treated as irreversibly anonymous, but the audit finding here is data-quality gates do not validate semantic feature definitions across pipeline versions. The relevant measure is “Introduce contract tests for unit, range and meaning changes before retraining”.
Question 197
In an audit of Data Management Specific to AI, managers say this measure is live: Introduce contract tests for unit, range and meaning changes before retraining. They claim it corrected a material audit gap. What direct operating evidence would allow the auditor to substantiate that particular claim across a defensible sample?
- Source telemetry and imputation trace on known outage intervals
- Schema diff, unit conversion assertions and rejected malformed batches
- Versioned data manifests, build IDs and successful sample reconstruction
- Reidentification tests, key access records and release-risk signoff
Explanation
Correct Answer
B:Correct. The proposed records—schema diff, unit conversion assertions and rejected malformed batches—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is data-quality gates do not validate semantic feature definitions across pipeline versions.
Incorrect Answers
A:“Source telemetry and imputation trace on known outage intervals” would test imputation logic confuses sensor outages with genuine zero measurements, not the present safeguard against data-quality gates do not validate semantic feature definitions across pipeline versions. The expected audit evidence is “Schema diff, unit conversion assertions and rejected malformed batches”.
C:“Versioned data manifests, build IDs and successful sample reconstruction” would test data reproducibility and provenance are absent from the ML pipeline, not the present safeguard against data-quality gates do not validate semantic feature definitions across pipeline versions. The expected audit evidence is “Schema diff, unit conversion assertions and rejected malformed batches”.
D:“Reidentification tests, key access records and release-risk signoff” would test pseudonymization was treated as irreversibly anonymous, not the present safeguard against data-quality gates do not validate semantic feature definitions across pipeline versions. The expected audit evidence is “Schema diff, unit conversion assertions and rejected malformed batches”.
Question 198
In a review of Data Management Specific to AI, investigators cannot establish the expected control proof: schema diff, unit conversion assertions and rejected malformed batches. The underlying event was that the data owner signs off on records before a schema change that silently converts currency units. Which control weakness is the MOST likely cause requiring focused audit work?
- Data-quality gates do not validate semantic feature definitions across pipeline versions
- Imputation logic confuses sensor outages with genuine zero measurements
- Data reproducibility and provenance are absent from the ML pipeline
- Pseudonymization was treated as irreversibly anonymous
Explanation
Correct Answer
A:Correct. The observed condition specifically indicates data-quality gates do not validate semantic feature definitions across pipeline versions. A meaningful audit response is to verify this diagnosis and then recommend introduce contract tests for unit, range and meaning changes before retraining.
Incorrect Answers
B:“Imputation logic confuses sensor outages with genuine zero measurements” concerns a separate audit weakness, not the missing assurance described by data-quality gates do not validate semantic feature definitions across pipeline versions. The correct cause is “Data-quality gates do not validate semantic feature definitions across pipeline versions”.
C:“Data reproducibility and provenance are absent from the ML pipeline” concerns a separate audit weakness, not the missing assurance described by data-quality gates do not validate semantic feature definitions across pipeline versions. The correct cause is “Data-quality gates do not validate semantic feature definitions across pipeline versions”.
D:“Pseudonymization was treated as irreversibly anonymous” concerns a separate audit weakness, not the missing assurance described by data-quality gates do not validate semantic feature definitions across pipeline versions. The correct cause is “Data-quality gates do not validate semantic feature definitions across pipeline versions”.
Question 199
A team can deploy a model directly from a notebook without peer review or signed lineage. From an AI audit perspective, which corrective action is the MOST direct response to the weakness described, rather than a useful control for some different AI risk?
- Obtain risk-proportionate independent validation before production authorization
- Rebaseline intended-use requirements and update safety, privacy and validation tests
- Rotate service credentials and transfer documented operations ownership at contract exit
- Require reproducible build, independent review and signed model registry promotion
Explanation
Correct Answer
D:Correct. Require reproducible build, independent review and signed model registry promotion is aimed at the actual failure: build and release provenance is bypassed by ad hoc notebook deployment. The audit should subsequently obtain promotion approvals, immutable build hashes and traced artifact lineage to test execution, not merely the existence of a policy.
Incorrect Answers
A:“Obtain risk-proportionate independent validation before production authorization” is a control for segregation of model development and challenge is missing, but the audit finding here is build and release provenance is bypassed by ad hoc notebook deployment. The relevant measure is “Require reproducible build, independent review and signed model registry promotion”.
B:“Rebaseline intended-use requirements and update safety, privacy and validation tests” is a control for requirements traceability failed across scope change, but the audit finding here is build and release provenance is bypassed by ad hoc notebook deployment. The relevant measure is “Require reproducible build, independent review and signed model registry promotion”.
C:“Rotate service credentials and transfer documented operations ownership at contract exit” is a control for knowledge transfer and ownership transition are absent from lifecycle controls, but the audit finding here is build and release provenance is bypassed by ad hoc notebook deployment. The relevant measure is “Require reproducible build, independent review and signed model registry promotion”.
Question 200
In an audit of AI Solution Development Methodologies and Lifecycle, managers say this measure is live: Require reproducible build, independent review and signed model registry promotion. They claim it corrected a material audit gap. What direct operating evidence would allow the auditor to substantiate that particular claim across a defensible sample?
- Separate validator signoff, challenge findings and closure evidence
- Signed requirements diff and refreshed trace from risks to test cases
- Promotion approvals, immutable build hashes and traced artifact lineage
- Offboarding checklist, credential revocation evidence and runbook acceptance
Explanation
Correct Answer
C:Correct. The proposed records—promotion approvals, immutable build hashes and traced artifact lineage—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is build and release provenance is bypassed by ad hoc notebook deployment.
Incorrect Answers
A:“Separate validator signoff, challenge findings and closure evidence” would test segregation of model development and challenge is missing, not the present safeguard against build and release provenance is bypassed by ad hoc notebook deployment. The expected audit evidence is “Promotion approvals, immutable build hashes and traced artifact lineage”.
B:“Signed requirements diff and refreshed trace from risks to test cases” would test requirements traceability failed across scope change, not the present safeguard against build and release provenance is bypassed by ad hoc notebook deployment. The expected audit evidence is “Promotion approvals, immutable build hashes and traced artifact lineage”.
D:“Offboarding checklist, credential revocation evidence and runbook acceptance” would test knowledge transfer and ownership transition are absent from lifecycle controls, not the present safeguard against build and release provenance is bypassed by ad hoc notebook deployment. The expected audit evidence is “Promotion approvals, immutable build hashes and traced artifact lineage”.