You apply for a job on a Monday morning. By Monday afternoon, a rejection arrives. No phone screen, no human review, no explanation. Just a system-generated email telling you the position has been filled or that you weren’t selected. It happens faster than any hiring manager could have read your resume, because no hiring manager did. An automated tool made the call.
What most New Jersey applicants don’t know is that this scenario is now squarely addressed by state law. On January 9, 2025, the New Jersey Attorney General and the Division on Civil Rights issued guidance confirming that automated decision-making tools used in hiring are covered by the New Jersey Law Against Discrimination. That means an algorithm can discriminate in the legal sense, and the employer can be held responsible for it. We represent only employees, and what follows is what New Jersey workers actually need to understand about these tools and their rights under current law.
How an AI Hiring Tool Can Discriminate Without a Human Ever Meaning To
Algorithmic discrimination doesn’t require a single biased thought from any person involved. Bias enters these tools in at least three distinct ways, and understanding the mechanism matters for recognizing what happened to you.
Discriminatory Design
The first point of failure is design. When a developer chooses what inputs the model uses, the criteria it weights, and the outcomes it optimizes for, those choices can build discrimination into the system before it processes a single application. The Equal Employment Opportunity Commission brought an action against a tutoring company whose screening tool was programmed to reject female applicants 55 and older and male applicants 60 and older. That was an intentional design choice, but many discriminatory design choices are subtler and produce the same result.
Biased Training Data
The second is training data. Many automated hiring tools learn to predict “good candidates” by studying an employer’s past hires. If those past hires skewed toward one race, one sex, or one age group because of historical discrimination, the model learns to replicate that skew. The tool isn’t programmed to exclude anyone; it’s programmed to find more people who look like the people who were hired before.
Flawed Deployment
The third is deployment. A tool validated for one purpose can be repurposed for something it was never tested against. A tool built for hourly roles may perform very differently when applied to salaried positions, and different standards applied to different applicant pools can create protected-class disparities on their own. Proxy variables compound this problem: a tool that screens by zip code, voice pattern, or response cadence can function as a stand-in for race, national origin, or disability without anyone explicitly programming it to do so.
New Jersey Has Confirmed the LAD Applies to Algorithmic Discrimination
The New Jersey Law Against Discrimination has prohibited employment discrimination based on race, sex, age, disability, national origin, and other protected characteristics for decades. What the January 9, 2025 guidance made explicit is that the law applies to automated tools the same way it applies to human decision-makers. The same three legal theories available in traditional discrimination cases apply when the decision was made by an algorithm: disparate treatment (the tool was used differently based on protected characteristics, treating one group worse than another), disparate impact (the tool, regardless of intent, produced outcomes that fell more heavily on a protected class without a legitimate business justification), and failure to accommodate (when an automated system has no mechanism for an applicant with a disability to request or receive a reasonable adjustment to the process).
The guidance also addresses a gap that employers have tried to use as a shield: the fact that a discriminatory tool was built by a third-party vendor. Under the LAD, an employer remains liable for algorithmic discrimination even when they didn’t design or build the tool themselves. Purchasing or licensing a system that produces unlawful outcomes doesn’t transfer the legal responsibility away from the company using it.
On December 15, 2025, the Division on Civil Rights published disparate impact rules at N.J.A.C. 13:16 in the New Jersey Register. These rules go further than the January guidance by specifically addressing how automated online application tools and facial analysis software used in hiring can generate disparate impact liability. Facial analysis software, used in some video-interview platforms to score candidates on inferred traits, is now explicitly within scope. New Jersey’s regulatory framework is more specific on this point than most other states have been.
You Don’t Need to Prove Intent to Have a Claim
This is the part of the law that surprises most workers. Many people assume that discrimination requires a biased person making a biased decision. It doesn’t. A claim based on disparate impact requires no proof of intent at all. If a facially neutral tool produces outcomes that fall disproportionately on applicants of a particular race, sex, age group, or disability status, and the employer can’t justify that disparity as a business necessity, that can be enough to support a claim under the LAD.
The practical challenge is recognition. New Jersey doesn’t currently require employers to disclose when an automated tool was used in hiring. An applicant may receive a rejection that looks like any other and never suspect a resume-screening algorithm, a video-interview scoring system, or a personality-assessment tool played any role. In a traditional hiring dispute, the discriminatory act usually involves some human contact that gives the applicant a sense that something was wrong. With automated screening, the rejection arrives with nothing to examine.
This makes documentation more important, not less. Save the exact date and content of any automated rejection email. Keep the original job posting, including any language about assessments, video interviews, or screening steps. If a platform required you to complete a video interview or online assessment, note that in writing and preserve any confirmation emails. These details establish the timeline and may later help identify which tools were in use and whether a pattern of outcomes exists.
Filing a Claim: DCR Complaint or Superior Court Lawsuit
Two paths exist for pursuing a LAD claim in New Jersey, and the deadlines are strict.
Administrative Complaint with the DCR
An administrative complaint with the New Jersey Division on Civil Rights must be filed within 180 days of the discriminatory act. That clock starts on the date of the automated rejection, not the date you later learned a tool was involved. 180 days passes faster than most people expect, particularly when weeks go by waiting to hear back about another application or hoping the first employer made a mistake.
Lawsuit in New Jersey Superior Court
The second option is filing a lawsuit directly in the Law Division of the New Jersey Superior Court. That deadline is two years from the discriminatory act. This path generally allows for broader discovery, including the possibility of seeking information about how a vendor’s tool was designed and validated, which can be critical when the discrimination is embedded in an algorithm rather than a conversation.
Choosing Between the Two Paths
One constraint applies regardless of which path you choose: you can’t pursue the same claim in both forums at the same time. A claim pending before the DCR can’t simultaneously be the subject of a Superior Court lawsuit. More consequentially, if the DCR issues a finding of no probable cause, that finding can foreclose a separate court filing on the same facts. Choosing between these paths isn’t a procedural formality. It’s a strategic decision that affects what evidence gets developed, how long the process takes, and what remedies are available.
What to Do If You Suspect an AI Tool Was Involved in Your Rejection
Start by writing down everything you remember about the application process while it’s still fresh. Note how quickly the rejection arrived, whether any automated steps were required, and what the job posting described about the selection process. Some platforms name their assessment or interview vendors directly in their terms of service or privacy disclosures; reviewing those can sometimes confirm which tools were used.
If the rejection came within minutes or hours of a completed application, that timing alone is worth preserving as a note. A human reviewer working through a full application stack almost never moves that fast. Speed isn’t proof of discrimination, but it’s a signal that automated screening was likely involved, and that the 180-day clock may already be running.
New Jersey law now treats an algorithm’s hiring decision the same as a person’s. If you have questions about whether an automated rejection violated the LAD and how much time you have to act, Green Savits represents only employees in New Jersey employment disputes. Reach out to us at (973) 965-8571.