Describe a scenario where an AI model's decision might introduce bias or risk, and how would you mitigate it?
