IAPP Updated AIGP Exam Questions and Answers by zack

ISO/IEC 42001:2023 establishes an Artificial Intelligence Management System designed to integrate AI governance into an organization ' s existing management structure rather than create an isolated governance regime. Its AI-policy requirements state that the policy should refer, where relevant, to other organizational policies. Therefore, A most accurately describes the relationship. The standard does not require the organization to absorb all related security, privacy, quality, risk, or compliance policies into one AI policy, making B incorrect. C is inconsistent with integrated governance because AI issues commonly intersect with information security, privacy, procurement, data governance, human resources, and enterprise risk management. Management review is an important management-system activity, but D does not state the relationship required between the AI policy and existing policies. ISO/IEC 42001 consequently favors coordinated and mutually aligned organizational governance.

================

Model disgorgement is the technique used to remove the effects of improperly used data from an ML system. This process involves retraining or adjusting the model to eliminate any biases or inaccuracies introduced by the inappropriate data. It ensures that the model ' s outputs are not influenced by data that was not meant to be used or was used incorrectly. Reference: AIGP Body of Knowledge on Data Management and Model Integrity.

Retraining an LLM (Large Language Model) is primarily done to improve or maintain its performance as data changes over time, to fine-tune it for specific use cases, and to incorporate new data interpretations to enhance accuracy and relevance. However, ensuring interpretability of the model ' s predictions is not typically a reason for retraining. Interpretability relates to how easily the outputs of the model can be understood and explained, which is generally addressed through different techniques or methods rather than through the retraining process itself. References to this can be found in the IAPP AIGP Body of Knowledge discussing model retraining and interpretability as separate concepts.

The correct answer is D because one of the most significant risks associated with generative AI is the potential leakage of sensitive, confidential, or proprietary information. AI governance frameworks emphasize data protection, confidentiality, and secure data handling as critical organizational responsibilities. When employees input sensitive data into generative AI tools, especially external or third-party systems, that information may be stored, reused, or exposed unintentionally. This creates legal, regulatory, and reputational risks. As a result, organizations may restrict or prohibit generative AI use to prevent unauthorized data disclosure. The other options relate to operational or business considerations, which are not primary governance drivers. Protecting sensitive information aligns with core AI governance principles such as privacy, security, and risk mitigation, making it the strongest justification for such a policy.