AI Regulation Tracker  /  Rule adopted

New Jersey Extends Disparate-Impact Rules to AI Hiring Tools

New Jersey's Division on Civil Rights has adopted binding rules, N.J.A.C. 13:16, that codify disparate-impact liability under the state's Law Against Discrimination and spell out how that liability applies to artificial intelligence and other automated decision-making tools used in hiring. The rules took effect December 15, 2025. This is an enforceable regulation, not guidance.

By Anthony Guerriero  ·  The Leveraged Years Regulatory Research  ·  Published July 16, 2026  ·  Last updated July 16, 2026

On December 17, 2025, New Jersey Attorney General Matthew J. Platkin and the Division on Civil Rights announced that the DCR had adopted new rules codifying the prohibition on disparate-impact discrimination under the Law Against Discrimination. The rules sit at N.J.A.C. 13:16. They were adopted by DCR Director Yolanda N. Melville on November 5, 2025, filed the next day as R.2025 d.150, and published in the New Jersey Register on December 15, 2025, with an effective date of the same day.

The rules matter well beyond New Jersey because of one thing they do plainly: they explain how disparate-impact liability applies to artificial intelligence and other automated decision-making tools in employment. Most of the AI-hiring conversation to date has lived in guidance documents, advisories, and enforcement statements. New Jersey put it in an adopted rule.

What the rule actually does

Start with disparate impact, because that is the engine here. Under the LAD, a practice can be unlawful if it disproportionately harms members of a protected class, even when the employer had no intent to discriminate. The new rules codify that standard. As DCR Director Melville put it, "The LAD safeguards against not only intentional acts of discrimination but also practices and policies that disproportionately affect members of protected groups."

The rules reach broadly across the areas the LAD already governs: employment, housing, places of public accommodation, financial lending, and contracting. The DCR describes them as among the most comprehensive state-level disparate-impact regulations in the country. But the piece that makes this an AI story is narrower and specific. The rules explain, in the DCR's own words, "how disparate impact liability applies to the use of artificial intelligence (AI) and other automated decision-making tools in the employment context."

In practice, that means the automated tools employers now lean on in hiring are analyzed the same way any other screening practice would be. The DCR names the kinds of tools it has in mind: resume-screening algorithms, automated online application technology that filters applicants, and facial-analysis or video-based assessment tools. If one of those tools produces a discriminatory outcome against a protected group, the fact that a machine generated the result is not a defense.

Intent is not the test

This is the part employers most often get wrong, so it is worth being blunt. Disparate impact does not turn on whether anyone meant to discriminate. It turns on the effect. An AI hiring tool trained on a company's existing workforce can learn to favor candidates who look like the people already there. If that workforce skews one way, the tool can quietly reproduce that skew and screen out qualified applicants from protected groups. No one has to have programmed bias in for the outcome to be unlawful under the LAD.

The rules also set up how these claims get resolved. A complainant has to show the practice produces a disparate impact. The burden then shifts to the covered entity to show the practice is necessary to achieve a substantial, legitimate, nondiscriminatory interest, and that there is no less discriminatory, equally effective alternative that would serve the same interest. That framework applies to an algorithm exactly as it applies to a human process.

You cannot outsource the liability

One point deserves its own line because it catches employers off guard. Buying the tool from a vendor does not move the risk off your books. Where a covered entity's practice relies on a third-party product that produces a disparate impact, the entity is expected to take reasonable steps to confirm the vendor's product complies with the LAD and the rules. "We used a vendor" is not an answer. If the tool screens people out, the employer using it owns that outcome.

What this is, and what it is not

Be precise about the scope. This is a binding New Jersey rule. It has the force of law inside New Jersey, and it took effect on December 15, 2025. It is not federal law, and it does not directly bind employers in other states.

It is also worth being precise about the AI-specific reach. The disparate-impact standard in the rules spans employment, housing, public accommodations, lending, and contracting. The DCR's explanation of how the standard applies specifically to AI and automated decision tools is framed in the employment context. That is the ceiling of what the agency has spelled out for algorithms, and it is where an employer using AI hiring tools should focus. Do not read the rule as a broad AI statute covering every sector's use of automation. Read it as a civil-rights rule that, among other things, makes clear that automated hiring tools are not exempt from disparate-impact liability.

Attorney General Platkin framed the adoption in civil-rights terms, saying, "The landmark rules we are adopting today further solidify our state's nation-leading civil rights protections." The signal to employers is that the state sees AI hiring tools squarely inside that protection, not outside it.

What this means for employers and their advisers

If you hire in New Jersey and you use any automated tool to screen, rank, or filter applicants, this rule already applies to you. The tool is now subject to disparate-impact analysis, and the burden of justifying it, if it produces skewed outcomes, sits with you. The practical response is not to abandon the tools. It is to be able to show your work.

For employers outside New Jersey, treat this as an early read on where state civil-rights enforcement is going. New Jersey is among the first states to write the AI-hiring principle into a binding rule instead of a guidance letter, and other states watch each other. The controls that make you defensible in New Jersey are the same controls that will hold up wherever this idea spreads: know what your tool is doing, test it for disparate impact, keep the evidence, and do not assume a vendor's assurances protect you.

The through-line is documentation. The rule's burden-shifting framework rewards employers who can point to a real, tested justification for a screening practice and punishes the ones who cannot explain why their tool rejects who it rejects. If you cannot answer why the algorithm screened someone out, you are not ready for a disparate-impact claim.

What to do now

Inventory every automated tool touching your New Jersey hiring, from resume screeners to scheduling filters to video assessments. Ask your vendors, in writing, how their product has been tested for disparate impact across protected classes, and keep the answers. Run your own outcome testing rather than relying on the vendor's marketing. Document a substantial, legitimate, nondiscriminatory business reason for any screen you rely on, and check whether a less discriminatory, equally effective alternative exists. And talk to employment counsel licensed in New Jersey before you change a hiring practice, because the rule is binding and the burden, once impact is shown, is yours.

Browse the full AI Regulation News tracker

Informational analysis for working professionals, not legal advice. Confirm how any rule or requirement applies to your situation with qualified counsel in the relevant jurisdiction.