An IS auditor is evaluating controls for detecting biases and attacks in an AI system during its operation.Which of the following approaches BEST ensures identification of anomalies affecting the model’s performance?
Answer(s): B
Real-time monitoring with automated detection and alerts allows continuous identification of anomalies, bias drift, or performance degradation during operation, providing immediate visibility and enabling timely corrective actions to maintain model reliability and integrity.
A healthcare provider uses a machine learning (ML) model with patient medical records and is concerned that attackers could reconstruct patient information through crafted queries to the model.Which of the following is the MOST important consideration when evaluating this privacy risk?
Differential privacy techniques protect individual records from being inferred through model queries by introducing controlled noise, directly mitigating the risk of reconstruction attacks and ensuring that sensitive patient data remains confidential while enabling model utility.
An organization has deployed a neural network-based AI system. An IS auditor discovers prompts can manipulate system outputs, potentially causing misleading results.Which of the following is the auditor’s BEST recommendation?
Answer(s): D
Rigorous input validation and anomaly detection prevent malicious or unexpected prompts from producing misleading outputs, ensuring that system responses remain reliable and that manipulation attempts are detected and mitigated before impacting decision-making or operational integrity.
Which of the following techniques BEST allows the AI system to detect rare events and improve training data issues?
Answer(s): C
Oversampling minority classes increases the representation of rare events in the training data, improving the model’s ability to learn from these cases and enhancing detection accuracy, thereby addressing class imbalance and mitigating bias in model predictions.
A real estate agency uses an AI tool to predict house values. The model accuracy significantly declined after six months in production.Which of the following is the MOST likely root cause?
Covariate shift occurs when the statistical properties of input data change over time compared with the training data, causing model performance to degrade in production, which explains the decline in predictive accuracy observed after six months.
Post your Comments and Discuss ISACA AAIA exam prep with other Community members:
Our website is free, but we have to fight against AI bots and content theft. We're sorry for the inconvenience caused by these security measures. You can access the rest of the AAIA content, but please register or login to continue.