AAISM 試験問題を無料オンラインアクセス

An organization is designing an AI-based credit risk assessment system that will integrate with sensitive financial datasets. Which of the following would BEST support the implementation of security-by-design principles in the AI system's architecture?

• A.Applying threat modeling specific to AI components before deployment
• B.Restricting access to AI models using IP allow lists to reduce public exposure
• C.Segmenting AI services across containers to manage resource constraints
• D.Integrating differential privacy mechanisms into model training to limit data leakage

Which of the following BEST supports the evaluation of AI system outputs?

• A.Input controls
• B.Scenario analysis
• C.Manual review
• D.Model card assessment

Which of the following strategies is the MOST effective way to protect against AI data poisoning?

• A.Ensuring the model is trained on diverse data sources
• B.Using robust data validation techniques and anomaly detection
• C.Increasing model complexity
• D.Incorporating more features and data into model training

After deployment, an AI model's output begins to drift outside of the expected range. Which of the following is the development team's BEST course of action?

• A.Create an emergency change request to correct the issue
• B.Adjust the hyperparameters of the AI model
• C.Take the AI model offline
• D.Return to an earlier phase in the AI life cycle

Which of the following is the MOST effective way to identify deepfakes generated by AI?

• A.Examining irregularities in metadata or data traces
• B.Detecting irregular pixel patterns or distorted frames
• C.Comparing the model card to the output generated by AI
• D.Analyzing biometric inconsistencies