Recognize AI-driven Threats CY0-001 Practice Questions (Page 6)

Questions 26–30

1. 26foundation · medium

   How can AI be used to enhance malware capabilities?

   • ABy making malware larger in size.
   • BBy enabling malware to adapt its behavior to evade detection.
   • CBy ensuring malware only runs on specific hardware.
   • DBy making malware visible to security tools.
   Reveal Solution Discussion Select an answer first
2. 27expert · hard

   A security team has deployed an AI-based malware detection system that uses a deep learning model. The team notices that the system is failing to detect certain malware samples that were previously detected. After investigation, they find that the malware samples have been modified with small, imperceptible changes that cause the model to classify them as benign. Which type of attack has occurred?

   • AModel inversion
   • BAI-driven phishing
   • CData poisoning
   • DAdversarial AI
   Reveal Solution Discussion Select an answer first
4. 28application · medium

   An organization notices a spike in phishing emails that are highly personalized, using the recipient's recent purchases and social media activity to craft convincing messages. The emails are sent at scale and adapt their content based on whether the recipient clicks a link. Which AI-driven threat is being used?

   • AData poisoning
   • BModel inversion
   • CEvasion attack
   • DAI-driven phishing
   Reveal Solution Discussion Select an answer first
5. 29foundation · medium

   Which stage of the AI lifecycle is most directly targeted by a data poisoning attack?

   • AInference
   • BDeployment
   • CTraining
   • DMonitoring
   Reveal Solution Discussion Select an answer first
6. 30application · medium

   A bank uses an AI model to approve credit card transactions. An attacker wants to cause the model to approve fraudulent transactions. The attacker gains access to the model's training data and inserts a small number of fraudulent transactions labeled as legitimate. What is the likely outcome?

   • AThe model will be unaffected because the number of samples is small
   • BThe model will learn to classify fraudulent transactions as legitimate
   • CThe model will reject all transactions to be safe
   • DThe model will become more accurate at detecting fraud
   Reveal Solution Discussion Select an answer first