Free ISACA AAIR Exam Dumps Questions & Answers

An organization is launching multiple generative AI use cases. The risk practitioner needs to ensure that AI risk decisions align with business objectives and approved risk boundaries of the enterprise.
What is the best action to take?

• A.Include AI in risk appetite and tolerance.
• B.Develop an AI-specific risk register and control roadmap.
• C.Increase investment in data integrity controls.
• D.Adopt a reactive risk management approach.

Which of the following is a risk practitioner ' s BEST recommendation to establish accountability for AI system outputs and decisions?

• A.Regular reviews of resource allocation for AI projects
• B.Continuous monitoring and key performance indicators (KPIs)
• C.Centralized governance task force for model decision authority
• D.Formal documented role assignments with named owners

Which of the following is the MOST important consideration when assessing risk for an AI system provided by a third party, as opposed to a non-AI related system?

• A.More stringent authentication requirements for access to core AI system functionalities
• B.Reduced reliance on structured frameworks or methodologies to assess impact and likelihood
• C.The use of quantitative data as a primary input to inform risk assessments
• D.Greater emphasis on continuous monitoring due to the dynamically evolving nature of AI models

Which of the following should be the PRIMARY consideration when determining the priority for restoration of AI systems following a model exfiltration attack?

• A.Reliance on the AI system for critical business requirements
• B.AI-specific expertise among business continuity team members
• C.Availability of externally vetted datasets for AI model retraining
• D.Cost of AI system vulnerability testing and patch deployment

Which of the following is the PRIMARY benefit of retraining an AI model with refreshed data?

• A.Incorporating stakeholder feedback to optimize model decisions and outputs
• B.Reducing errors and misclassifications by allowing the model to learn from recent examples
• C.Enabling more precise tuning of model hyperparameters to improve performance
• D.Allowing the removal of outdated personal information contained in datasets