AI Risk and Threat Management CRAGE Practice Questions (Page 10)

Questions 46–50

1. 46foundation · easy

   Which activity is part of a periodic AI risk review process?

   • ADeleting the risk register after each review
   • BFreezing all AI system changes permanently
   • CIgnoring new risks that emerge after deployment
   • DRe-evaluating the likelihood and impact of previously identified risks
   Reveal Solution Discussion Select an answer first
2. 47application · medium

   A company is considering using a third-party vendor's AI model for fraud detection. The vendor has provided limited documentation about the model's training data and performance. The company's risk team needs to assess the risks of using this third-party AI. What should the risk team do?

   • AConduct a third-party AI risk assessment that includes requesting additional documentation, testing the model, and reviewing the vendor's security practices.
   • BRequire the vendor to sign a contract that transfers all liability to the vendor.
   • CDeploy the model immediately and monitor its performance after launch.
   • DAccept the vendor's claims of model performance without further investigation.
   Reveal Solution Discussion Select an answer first
4. 48foundation · easy

   Which activity is part of assessing AI risks introduced by a third-party vendor?

   • AReviewing the vendor's AI model documentation and testing results
   • BDelegating all risk management to the vendor without oversight
   • CAssuming the vendor's AI system is risk-free because it is externally developed
   • DIgnoring the vendor's data handling practices
   Reveal Solution Discussion Select an answer first
5. 49application · medium

   A social media platform uses an AI content moderation system. The risk team is identifying potential risks associated with the system. Which of the following is an example of a societal risk that should be categorized?

   • AThe model could inadvertently suppress freedom of speech by over-flagging legitimate content.
   • BThe model's inference latency could increase during peak usage hours.
   • CThe training data could contain duplicate entries that reduce model accuracy.
   • DThe model's API could be accessed by unauthorized users.
   Reveal Solution Discussion Select an answer first
6. 50foundation · easy

   In AI threat modeling, which threat involves an attacker inserting malicious data into the training set to corrupt the model's behavior?

   • ADenial of service
   • BModel inversion
   • CData poisoning
   • DPrompt injection
   Reveal Solution Discussion Select an answer first