Artificial Intelligence Fundamentals GASAE Practice Questions (Page 7)

Questions 31–35

1. 31foundation · easy

   Which metric is commonly used to evaluate the effectiveness of an AI security tool in terms of false positives?

   • ARecall
   • BPrecision
   • CThroughput
   • DLatency
   Reveal Solution Discussion Select an answer first
2. 32application · medium

   A red team is planning an adversary emulation exercise against an AI-driven endpoint detection and response (EDR) platform. The goal is to test whether the EDR's AI model can be evaded by an attacker who has access to the model's architecture and weights. Which attack technique is most appropriate?

   • ASocial engineering of the EDR vendor's support staff
   • BBlack-box evasion using a substitute model
   • CData poisoning by injecting malicious samples into the training set
   • DWhite-box adversarial attack, crafting inputs using gradient information
   Reveal Solution Discussion Select an answer first
4. 33application · medium

   A security engineer wants to build a system that automatically categorizes network traffic as normal or suspicious. The engineer has a large dataset of labeled traffic samples. Which machine learning approach is most appropriate for this task?

   • AUnsupervised clustering to group traffic without labels
   • BReinforcement learning to maximize a reward signal
   • CA rule-based expert system with manually defined thresholds
   • DSupervised learning with a neural network classifier
   Reveal Solution Discussion Select an answer first
5. 34foundation · easy

   Which statement best describes the relationship between machine learning and security automation?

   • AMachine learning models can be trained to recognize patterns indicative of threats, enabling automated detection and response.
   • BMachine learning requires labeled data for every security scenario and cannot work with unlabeled data.
   • CMachine learning is only useful for post-incident forensics, not real-time detection.
   • DMachine learning eliminates the need for security policies and human oversight.
   Reveal Solution Discussion Select an answer first
6. 35application · medium

   During an adversary emulation exercise, a red team successfully crafts a malicious file that the AI-based antivirus classifies as benign. The team did not have access to the model's internals but used a separate model trained on similar data to generate the file. Which attack technique did the red team employ?

   • AModel inversion to extract training data
   • BWhite-box gradient-based attack
   • CData poisoning of the target model's training set
   • DTransferability attack using a substitute model
   Reveal Solution Discussion Select an answer first