A risk manager reviewing an AI-powered insurance underwriting system notices that the model assigns lower premiums to customers whose web browsing activity suggests interest in healthy lifestyle content. Which risk does this MOST directly raise?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — pricing off web browsing that hints at healthy lifestyle can become proxy discrimination, because you're using inferred sensitive attributes to set premiums. That's a legal and ethical risk, not just a model-quality issue.
Full Explanation
The correct risk is use of inferred sensitive attributes as proxy variables in pricing decisions, because web browsing activity suggesting interest in healthy lifestyle content may be used as a proxy for protected health status or other sensitive attributes, creating legal and ethical risk. Model underfitting due to insufficient feature engineering is a model quality issue, not the primary concern here. Data integrity risk from web scraping unreliable sources is a data sourcing risk but not the primary concern here. Operational risk from API latency in real-time decisioning is an infrastructure concern.