Security Controls and Monitoring AAISM Practice Questions (Page 4)

Questions 16–20

1. 16expert · hard

   A multinational corporation operates AI systems in multiple regions, each with different data protection regulations. The corporation must ensure that its AI security controls meet the most stringent regulatory requirements across all regions. What is the most effective strategy?

   • AApply the controls of the region with the least stringent requirements to all regions.
   • BImplement a unified set of AI security controls that meet the highest regulatory standard across all regions.
   • CDevelop separate control frameworks for each region to minimize costs.
   • DOutsource AI security monitoring to a third-party vendor that claims to be compliant with all regulations.
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   Which security control framework is specifically designed to provide a comprehensive set of controls for AI systems, addressing areas such as data governance, model development, and operational monitoring?

   • ANIST Cybersecurity Framework (CSF)
   • BISO/IEC 27001
   • CCOBIT 2019
   • DNIST AI Risk Management Framework (AI RMF)
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   An e-commerce company uses a recommendation engine that has been performing well for months. Recently, the click-through rate has declined significantly, and the recommendations seem less relevant. The data science team suspects model drift. What is the most appropriate first step to address this?

   • AImplement continuous monitoring of the model's performance metrics and input data distribution.
   • BIncrease the frequency of model retraining to daily.
   • CRetrain the model immediately with the most recent data.
   • DRoll back to a previous version of the model.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   Which attack involves an attacker intentionally providing malicious training data to an AI model to cause it to learn incorrect behaviors or classifications?

   • AData poisoning
   • BModel inversion
   • CAdversarial perturbation
   • DPrompt injection
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A company is implementing AI security controls and wants to ensure they align with a framework that is widely used in the industry. The company's security team is familiar with the NIST Cybersecurity Framework. What is the most appropriate way to apply this framework to AI security?

   • AReplace the NIST CSF with the OWASP Top 10 for LLMs.
   • BUse the NIST CSF only for the IT infrastructure supporting AI, not the AI models themselves.
   • CMap AI-specific controls to the NIST CSF functions: Identify, Protect, Detect, Respond, Recover.
   • DApply the NIST CSF without any modifications to AI systems.
   Reveal Solution Discussion Select an answer first