A Rule With National Reach
New Jersey's Division on Civil Rights adopted rules in December 2025, effective December 15, that its own attorney general describes as the most comprehensive state-level disparate impact regulations in the country. They apply to housing, lending, employment, public accommodations and contracting, and they reach into the newest frontier of fair lending: liability from artificial intelligence and automated decision-making tools.
The Mortgage Bankers Association filed suit in federal court on September 3 to block the lending provisions, naming Attorney General Jennifer Davenport and the division's director, Yolanda Melville, as defendants. The association, which represents more than 2,000 members including more than 60 in New Jersey, is represented without charge by the Pacific Legal Foundation. No judge has ruled, and the state's allegations and the association's arguments remain untested in court.
The Two Legal Arguments
The complaint rests on two pillars. The first is constitutional. The MBA argues the rule discards the safeguards the Supreme Court built into disparate impact law in its 2015 decision in Texas Department of Housing and Community Affairs v. Inclusive Communities Project, which required a robust causation showing, room for businesses to justify legitimate policies, and a burden on challengers to prove a less discriminatory alternative exists. New Jersey's version, the suit says, reverses that burden and leaves race-conscious policymaking as the only certain way to avoid liability, which the association calls unconstitutional racial balancing.
The second is preemption. Federal fair lending law, the Fair Housing Act, the Equal Credit Opportunity Act and Regulation B, prohibit race-conscious credit decisions. If New Jersey's rule effectively requires lenders to weigh race to avoid liability, the MBA argues, simultaneous compliance with both regimes becomes impossible, and the federal scheme should displace the state one.
The state's position, as its officials have articulated it, is that the rule simply operationalizes New Jersey's own Law Against Discrimination, that disparate impact doctrine is settled law at both the state and federal level, and that the rule gives businesses guidance rather than mandates. The litigation will test whether the guidance functions as a mandate in practice.
The Specific Provisions Under Attack
The complaint's objections are concrete. The rule permits challenges to policies that have not yet been implemented, reversing the usual sequence in which a policy causes an outcome before it is challenged. It places the burden on businesses to prove no less discriminatory alternative exists, complete with empirical evidence. It allows diversity or increased access for underrepresented groups to justify a challenged practice, which the MBA reads as an invitation to race-conscious design. And it allows disparity to be shown using national census or survey data rather than a lender's own applicant pool, without requiring the disparity to be substantial or statistically significant.
MBA president Bob Broeksmit summarized the industry's position as lenders being put in the untenable position of having to prove a negative. The association says it is acting to prevent regulatory overreach.
What the Case Means for Lending
The stakes extend beyond New Jersey's borders because the rule addresses the question every lender is now asking about its own models: how do you prove an algorithm does not discriminate? Under the federal framework, a lender must show its model has a legitimate business justification and no less discriminatory alternative. Under the New Jersey framework as the MBA reads it, the lender must also demonstrate the absence of statistical disparity against standards that are difficult to define in advance, using data the lender does not control.
If the rule survives, other states with aggressive civil rights enforcement will copy it, and automated underwriting models will be built to its specifications by default, because the largest lenders operate nationally. If the rule falls on preemption grounds, the states lose a tool in exactly the area, AI-driven credit decisions, where the federal agencies have been slowest to act.
Both sides have something real to protect, which is why the case is worth watching on its merits rather than its politics. The rule is a serious attempt to update civil rights enforcement for automated lending. The lawsuit is a serious attempt to keep federal fair lending law as the single standard lenders must satisfy. A court will now decide whether the two can coexist.
Primary sources
- National Mortgage News on the filing for the complaint's structure and the rule's provisions.
- Pacific Legal Foundation case page for the parties and the constitutional arguments.
- Lexology summary of the challenge for the preemption analysis.