CAS-005 · Question #218
A security engineer is performing threat modeling for an AI training architecture. The architecture implements a CI/CD pipeline to train a new AI model on a fixed schedule with live data from a…
The correct answer is A. Identify trust boundaries. When threat modeling an AI training architecture with a CI/CD pipeline, the most appropriate action to focus on threat movement from development to production is to identify trust boundaries.
Question
A security engineer is performing threat modeling for an AI training architecture. The architecture implements a CI/CD pipeline to train a new AI model on a fixed schedule with live data from a back-end storage location. The engineer wants to use a threat modeling activity to focus on the threat as it moves through the CI/CD pipeline to the production environment. Which of the following is the most appropriate action for the engineer to take?
Options
- AIdentify trust boundaries.
- BExecute automated code reviews.
- CMap to OWASP Top 10.
- DDocument data flows.
How the community answered
(23 responses)- A83% (19)
- B4% (1)
- C4% (1)
- D9% (2)
Why each option
When threat modeling an AI training architecture with a CI/CD pipeline, the most appropriate action to focus on threat movement from development to production is to identify trust boundaries.
Identifying trust boundaries is a fundamental step in threat modeling, crucial for understanding where security perimeters exist and where control over data and execution shifts. This helps pinpoint potential vulnerabilities as threats might traverse different stages of the CI/CD pipeline and enter the production environment.
Executing automated code reviews is a specific security testing activity, not a high-level architectural threat modeling activity focused on the system's overall threat flow.
Mapping to OWASP Top 10 focuses on common web application vulnerabilities, which might not fully encompass the broader architectural and data flow threats present in an AI training CI/CD pipeline.
Documenting data flows is an essential prerequisite for threat modeling but is an information gathering step, not the primary threat modeling *activity* focused on analyzing threat movement across an architecture.
Concept tested: Threat modeling methodologies and trust boundaries
Source: https://learn.microsoft.com/en-us/security/engineering/overview-of-the-stride-model
Community Discussion
No community discussion yet for this question.