What is a common challenge in AI vendor management?

• A. Adapting existing vendor management programs to include AI-specific risks
• B. Reducing the number of suppliers regardless of risk
• C. Neglecting ethical considerations in procurement
• D. Outsourcing all risk to suppliers

Answer: (A)

AI vendor management hinges on updating existing programs to address AI-specific risks that traditional vendor controls don’t fully cover. AI systems bring data-centric and model-centric challenges that ripple through every stage of a vendor relationship. You need governance around data quality, provenance, privacy, and security, since data flows to and from vendors and often powers the models themselves. The model risk aspect is distinct from ordinary software risk: models can drift over time, produce biased or unreliable results, be brittle to inputs, or be vulnerable to adversarial manipulation. Because AI evolves with frequent updates and retraining, the governance framework must handle versioning, validation, ongoing monitoring, and clear accountability for decisions influenced by AI outputs. There’s also the need to manage the broader supply chain: data suppliers, model providers, and downstream integrators or subcontractors, plus the regulatory and ethical considerations that accompany AI deployment. In short, the common challenge is reshaping vendor management to explicitly cover AI-specific risks across data, models, and lifecycle.

Other options hint at important concerns, but they don’t capture this core shift. Simply trying to reduce suppliers ignores resilience and risk diversification, neglecting ethics is a separate governance issue, and outsourcing all risk is not a realistic or responsible stance.