More than once, the CFPB has paired novel analytical technology with an equally novel theory of discrimination while skipping the model-validation discipline federal regulators expect of supervised banks. This is the earlier and costliest instance: a BISG-based regression estimator that, under the disparate impact theory the agencies pled, roughly doubled the pricing disparities behind $160M+ in indirect-auto settlements—then became a default fair-lending test for industry co

The headline of the CFPB's new Regulation B is that ECOA no longer bars disparate impact. But for non-mortgage lenders, a quieter clause matters more: its intent-based proxy test is direction-agnostic, reaching variables used to advantage or disadvantage a protected class alike. That language captures exactly how modern algorithmically-debiased "less discriminatory alternative" credit models work—recasting a former compliance tool as potential disparate treatment.

The 2025 EO didn't kill disparate impact — SCOTUS settled that in Inclusive Communities. But agency enforcement overreach may have invited the blow.