Your AI Performance Review Tool Might Be Your New Bias Problem

HR teams spent two years hardening AI hiring tools against bias — but the same algorithmic logic now quietly runs inside performance review systems, promotion engines, and pay raise tools. And almost nobody is auditing it. In this episode, we break down a concept researchers are calling "bias laundering" — where managers selectively embrace AI-generated ratings when they confirm what they already want to believe, and quietly override the system when they don't. A UNSW study of 242 managers found a striking 18-point gap: 60% adopted algorithm-recommended high ratings, but only 42% used algorithm-recommended low ratings. The data looks objective. The bias is still there. We also walk through why this is a growing legal liability. The disparate impact doctrine doesn't stop at hiring — it applies to every employment decision, including promotions, raises, and PIPs. Fisher Phillips' 2026 analysis is clear: employers need to audit AI tools across the entire employee lifecycle, not just at the front door. Cases like Mobley v. Workday are already testing these theories in court. Finally, we share the five concrete steps HR leaders should take right now — starting with mapping every AI touchpoint in your employee lifecycle and demanding process transparency, not just output transparency. The EU AI Act is coming for post-hire systems too. The time to get ahead of this is before the regulation catches up.