Privacy-Enhancing Technologies CDPSE Practice Questions (Page 6)

Questions 26–30

1. 26foundation · easy

   Under privacy regulations like GDPR, what is generally required before setting non-essential cookies on a user's device?

   • AProviding a privacy policy that mentions cookies
   • BRegistering the website with a data protection authority
   • CAnonymizing the data collected by cookies
   • DObtaining the user's consent before the cookies are set
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   What is a model inversion attack in the context of AI and machine learning?

   • AAn attack that modifies the model's weights to change its behavior
   • BAn attack that floods the model with requests to cause denial of service
   • CAn attack that steals the model's source code
   • DAn attack that uses the model's outputs to reconstruct training data
   Reveal Solution Discussion Select an answer first
4. 28foundation · easy

   In pseudonymization, where is the mapping between original identifiers and pseudonyms typically stored?

   • AIn the same database as the pseudonymized data
   • BIt is not stored; the mapping is derived from the data itself
   • CIn the user's browser cookies
   • DIn a separate, securely controlled location
   Reveal Solution Discussion Select an answer first
5. 29application · medium

   A company's website uses a third-party analytics service that sets a persistent tracking cookie to measure user behavior across sessions. The company is based in the EU and must comply with the ePrivacy Directive. What is the primary legal requirement for setting this cookie?

   • AThe company must register the cookie with the data protection authority.
   • BThe company must include a privacy policy page that explains the use of cookies.
   • CThe cookie must be set to expire after 30 days to limit data collection.
   • DThe user must be provided with clear and comprehensive information and give their consent before the cookie is set.
   Reveal Solution Discussion Select an answer first
6. 30expert · hard

   A research institute is building a model to predict disease outbreaks using patient data from multiple hospitals. The data is highly sensitive, and the institute must comply with strict data protection regulations. The hospitals are willing to share data, but only if the risk of re-identification is minimized. The institute needs to train a model that is accurate enough for public health decisions. Which approach best addresses the need for both privacy and accuracy?

   • AUse federated learning to train the model across hospitals without sharing raw data.
   • BUse differential privacy with a high epsilon value to maintain model accuracy.
   • CUse homomorphic encryption to encrypt the data and train the model on encrypted data.
   • DCombine federated learning with differential privacy to train the model.
   Reveal Solution Discussion Select an answer first