This research introduces a scalable framework for assessing fairness in chatbots by focusing on "first-person fairness" - evaluating bias in direct user interactions rather than institutional decision-making.

• Develops counterfactual testing methods specifically designed for open-ended conversational AI systems
• Addresses the unique challenges of evaluating bias across diverse chatbot use cases from resume writing to entertainment
• Provides actionable approaches for detecting and mitigating harmful stereotypes in AI conversations
• Creates evaluation methods that accommodate the personalized, interactive nature of chatbot interactions

For security professionals, this research offers critical tools to identify and prevent biased outputs that could harm users or reinforce discrimination, helping organizations deploy more responsible AI systems.

First-Person Fairness in Chatbots