Expanding Robustness in Responsible AI for Novel Bias Mitigation