4.2 Define Secure AI and Its Importance in Protecting AI Systems from Malicious Attacks and Misuse. GENERATIVE-AI-LEADER Practice Questions (Page 3)

Questions 11–15

1. 11foundation · easy

   What is the primary purpose of Google's Secure AI Framework (SAIF)?

   • ATo automate the entire machine learning pipeline without human oversight
   • BTo provide a structured approach for securing AI systems against threats and misuse
   • CTo replace traditional network security tools with AI-based firewalls
   • DTo guarantee that AI models are 100% accurate on all inputs
   Reveal Solution Discussion Select an answer first
2. 12application · medium

   A company is using a gen AI model to generate code. They are concerned about the model being used to generate malicious code. They want to monitor the model's outputs to detect and block malicious code generation. Which monitoring approach is most effective?

   • AMonitor the model's resource usage to detect unusual activity.
   • BImplement output filtering and scanning to detect malicious patterns in the generated code.
   • CRestrict access to the model using IAM so only authorized users can generate code.
   • DDisable monitoring to avoid false positives.
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A financial services firm is deploying a gen AI application on Google Cloud. They are concerned about the underlying infrastructure being compromised, which could expose sensitive customer data processed by the AI. They want to leverage Google Cloud's foundational security capabilities to protect the AI workload. Which combination of measures best leverages Google Cloud's secure-by-design infrastructure?

   • AMove the AI workload to a single region to reduce the attack surface.
   • BApply IAM roles to the AI service accounts and rely on the principle of least privilege.
   • CUse Google Cloud's hardware root of trust, encrypted communication between services, and secure boot for the underlying VMs.
   • DEnable Security Command Center and review all findings daily.
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   A large enterprise is adopting generative AI across multiple business units. The CISO wants a structured approach to ensure AI systems are secure by design, not as an afterthought. They also need to align security practices across different teams that are building and deploying models independently. Which approach best addresses this need?

   • AAdopt Google's Secure AI Framework (SAIF) to provide a common set of security controls and best practices across the organization.
   • BDeploy a single centralized firewall to protect all AI workloads from external threats.
   • CRequire all teams to use the same pre-trained model to reduce the attack surface.
   • DConduct a one-time security audit of all existing AI projects and fix identified issues.
   Reveal Solution Discussion Select an answer first
6. 15expert · hard

   A company is adopting AI and wants to implement Google's Secure AI Framework (SAIF). They have a limited security team and need to prioritize their efforts. They are considering which SAIF principles to implement first. Which principle, if implemented first, would provide the most foundational protection for their AI systems?

   • AAlign security controls with business priorities and risk tolerance.
   • BAccelerate innovation within a secure framework by enabling rapid development with security guardrails.
   • CDevelop specialized defenses for AI/ML systems, such as adversarial input filtering.
   • DExpand strong security foundations to the AI ecosystem, including secure-by-design infrastructure and IAM controls.
   Reveal Solution Discussion Select an answer first