What describes vulnerabilities in any component relied upon by the AI system that can compromise the entire system?

• A. Supply chain vulnerabilities
• B. Model backdoor attack
• C. Response costs
• D. Fines/Penalties

Answer: (A)

Vulnerability in any component relied upon by the AI system describes supply chain vulnerabilities. This means weaknesses anywhere in the chain—the data sources, trained models, software libraries, hardware, firmware, cloud services, or development tools—that the AI depends on. If any of these elements are compromised, tampered, or subverted, the entire system’s integrity and function can be affected. Because the AI relies on multiple external and internal components, a flaw in one part can propagate and undermine the whole deployment.

A model backdoor attack, by contrast, is a specific malicious mechanism planted inside the model itself to elicit hidden behavior, not the broader category of vulnerabilities across all components. The other options refer to costs or penalties rather than the actual weakness in components that could jeopardize the whole system.