Threats and Vulnerabilities Specific to AI AAIA Practice Questions (Page 2)

Questions 6–10

1. 6expert · hard

   A company is building an AI system and wants to use a pre-trained model from an open-source repository. The company is concerned about supply chain risks. Which approach is most appropriate to mitigate these risks?

   • AThoroughly vet the model's provenance and license
   • BOnly use models from well-known vendors
   • CUse the model as-is to save time
   • DRetrain the model from scratch
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   What is the goal of a model inversion attack?

   • ATo create a copy of the model's architecture
   • BTo bypass the model's authentication controls
   • CTo degrade the model's performance
   • DTo reconstruct the training data used to build the model
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   What is the primary goal of a prompt injection attack on a large language model (LLM)?

   • ATo manipulate the model into performing unintended actions or revealing sensitive information
   • BTo steal the model's weights
   • CTo corrupt the model's training data
   • DTo overload the model with requests
   Reveal Solution Discussion Select an answer first
5. 9expert · hard

   A financial institution deploys a fraud detection model. The model is accurate but its decisions are not explainable, and regulators require explanations for denied transactions. The institution wants to maintain accuracy while meeting regulatory requirements. Which approach is most appropriate?

   • AUse a simpler, interpretable model even if accuracy drops slightly
   • BUse a complex model and provide no explanations
   • CUse a complex model and hire auditors to manually review decisions
   • DKeep the complex model and provide generic explanations
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   An AI development team uses an open-source library for data preprocessing. What is a potential AI supply chain vulnerability?

   • AThe library may be too slow for production use
   • BThe library may have a different license than expected
   • CThe library may not support the latest hardware
   • DThe library may contain malicious code that alters the training data
   Reveal Solution Discussion Select an answer first