3.2 Describe Approaches to Protect Corporate Data Privacy and Security in AI Systems AI-TECHNICAL-PRACTITIONER Practice Questions (Page 1)

Questions 1–5

1. 1foundation · easy

   Which organizational governance approach involves assigning a specific person or team to be responsible for overseeing data privacy and security practices within an AI project?

   • ADesignating a data protection officer (DPO)
   • BImplementing differential privacy
   • CConducting a data minimization review
   • DDeploying a firewall
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   A government agency is building an AI system to analyze citizen data for public policy. The agency must comply with strict data protection regulations. The data includes sensitive attributes such as income and health status. The agency wants to publish the model and its training statistics. The data science team proposes using differential privacy. The legal team is concerned that differential privacy may not be sufficient to protect against all re-identification attacks. Which additional measure should the agency implement?

   • ATrain the model using federated learning
   • BImplement a data governance framework that includes regular privacy impact assessments and access reviews
   • CRemove all sensitive attributes from the training data
   • DUse stronger encryption for the training data
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   Which data security measure would be most appropriate to protect sensitive personal data in an AI training dataset while still allowing the data to be used for statistical analysis?

   • AAccess controls
   • BAnonymization
   • CEncryption at rest
   • DData masking
   Reveal Solution Discussion Select an answer first
5. 4foundation · easy

   An organization trains an AI model on customer data collected for fraud detection. Later, it wants to use that same data for targeted advertising. Which data privacy principle would be most directly violated?

   • AStorage limitation
   • BData accuracy
   • CPurpose limitation
   • DData minimization
   Reveal Solution Discussion Select an answer first
6. 5expert · hard

   A research consortium is building a collaborative AI model using data from multiple pharmaceutical companies. Each company's data is proprietary and cannot be shared with other members. The consortium wants to train a model that benefits from all datasets while ensuring no company can infer another's proprietary data. Which technique should the consortium use?

   • AFederated learning with differential privacy on model updates
   • BHomomorphic encryption of all datasets before sharing
   • CData masking of all proprietary data before training
   • DCentralizing all data in a trusted third-party environment
   Reveal Solution Discussion Select an answer first