AI Risk Scenario Identification and Assessment (e.g., Threats, Vulnerabilities, and Attacks) AAIR Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   An organization has identified three AI risk scenarios with the following likelihood and impact ratings (on a scale of 1-5): Scenario 1: likelihood 5, impact 3; Scenario 2: likelihood 3, impact 5; Scenario 3: likelihood 4, impact 4. Which scenario should be prioritized first?

   • AAll scenarios equally
   • BScenario 3
   • CScenario 2
   • DScenario 1
   Reveal Solution Discussion Select an answer first
2. 12application · medium

   A risk analyst at a bank has identified two AI risk scenarios. Scenario A has a high likelihood and high impact. Scenario B has a low likelihood and high impact. The bank has limited resources to mitigate risks. Which scenario should be prioritized?

   • AScenario A, because it has a higher likelihood
   • BNeither scenario, because the bank has limited resources
   • CScenario B, because it has a high impact
   • DBoth scenarios equally, because they both have high impact
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A healthcare organization uses an AI system to triage patient radiology images. The system is hosted on a cloud platform and accessed by radiologists via a web portal. An attacker gains access to the system and is able to query the model with carefully crafted images to infer whether a specific patient's data was used in the training set. Which type of attack vector is being exploited?

   • AExtraction attack
   • BData poisoning attack
   • CEvasion attack
   • DInference attack
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   An e-commerce company uses a recommendation system to suggest products to customers. The system is trained on user behavior data. A competitor is suspected of submitting fake user accounts and artificially inflating the popularity of certain products to manipulate the recommendations. Which of the following risk scenarios BEST describes this situation?

   • AAn attacker gains unauthorized access to the model's deployment environment and alters its configuration.
   • BAn attacker poisons the training data by injecting fake user behavior to skew the model's recommendations.
   • CAn attacker extracts the model's parameters by querying it with carefully crafted inputs.
   • DAn attacker exploits a vulnerability in the model's architecture to cause misclassification.
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   When evaluating the impact of an AI risk scenario, which of the following should be considered?

   • AOnly the technical difficulty of the attack
   • BThe number of lines of code in the AI system
   • CThe model's accuracy on the training data
   • DEffect on organizational objectives and stakeholders
   Reveal Solution Discussion Select an answer first