top of page

Mobley v. Workday

What the AI Hiring Lawsuit Alleges and What the Court Has Decided

Alyssa I. Agard

Mobley v. Workday: What the AI Hiring Lawsuit Alleges and What the Court Has Decided explains the Mobley v. Workday, Inc. lawsuit over AI hiring tools as of October 26, 2026, and takes no position on the merits. It traces the case from Mobley’s 2021 EEOC charge to the plaintiffs’ September 2026 motion for class certification and sets out each side’s position and evidence. It stresses that the court’s rulings so far have applied early-stage legal standards and have not resolved any disputed facts. It then explains how Title VII, the ADEA, the ADA, and New Jersey’s Law Against Discrimination may apply when employers use a software vendor’s hiring tools, and it closes with guidance that official sources give New Jersey employers. Every claim is supported by footnotes to primary sources, including court filings, statutes, regulations, and official state guidance.

​

This analysis is for general information only. It describes allegations, court filings, and rulings as of October 26, 2026, and it is not legal advice. Agard Research Associates Inc. is not a law firm and takes no position on the merits of the case.

​

El resumen también están disponibles en español.

Support Our Research

Your contributions enable us to provide valuable insights and free resources to the community. Help us continue our mission by donating today.

Information current as of October 26, 2026

Summary

​

Four job seekers are suing Workday, Inc., a provider of human resource management services, in federal court in San Francisco, and more than 15,000 people have opted in to their age discrimination claim. They allege that Workday's artificial intelligence (AI) screening tools disproportionately exclude applicants who are Black, female, over 40, or disabled. Workday denies the allegations and contends that it is not covered by the federal anti-discrimination statutes the plaintiffs invoke. The court has allowed most of the plaintiffs' disparate impact claims to proceed past the pleading stage, at which a court accepts a complaint's factual allegations as true, and the plaintiffs' motion asking the court to certify a class lists a hearing date of March 9, 2027.

​

This analysis is for general information only. It describes allegations, court filings, and rulings as of October 26, 2026, and it is not legal advice. Agard Research Associates Inc. is not a law firm and takes no position on the merits of the case.

​

​

The Parties and the Platform

​

The case is Mobley v. Workday, Inc., No. 3:23-cv-00770-RFL, before Judge Rita F. Lin of the U.S. District Court for the Northern District of California. According to the operative complaint, the four named plaintiffs are Derek L. Mobley, a Black man over 40 with depression and anxiety; Jill E. Hughes, a woman over 40 whose disabilities include cancer and asthma; Sheilah Johnson-Rocha, a Black woman over 40; and FaithLinh Rowe, an Asian American woman over 40.

​

Workday, Inc., based in Pleasanton, California, offers human resource management services to medium and large organizations. Its answer admits that, "by at least one calculation, customers used Workday's software to process 2.2 million job requisition transactions in May 2023, representing nearly 22% of all U.S. job openings that month."

​

The plaintiffs' class certification motion challenges three tools: HiredScore Spotlight, Fetch, and Candidate Skills Match. HiredScore is a product that Workday acquired while the case was pending. HiredScore has described Spotlight as grading, on an A-to-D scale, how well an applicant's stated qualifications match a job's requirements, and Fetch as matching open jobs with earlier applicants who agreed to be considered for other positions at the company. The court's May 2025 order describes Candidate Skills Match as comparing the skills in an employer's job posting with an applicant's materials and reporting the result as "strong," "good," "fair," "low," "pending," or "unable to score." Workday has asserted, citing a declaration, that customers can "enable, disable, use, or ignore" the many features of its recruiting software, including its AI features.

​

​

​

How the Case Developed

​

June and July 2021

Mobley filed a discrimination charge, and then an amended charge, with the U.S. Equal Employment Opportunity Commission (EEOC). The amended charge alleged that Workday's selection tools disparately affect applicants because of race, age, and disability.

​

November 22, 2022

According to the complaint, the EEOC issued Mobley a "Dismissal and Notice of Right to Sue." Under the EEOC regulation on dismissals, which covers Title VII and ADA charges, that notice tells the recipient that a lawsuit may be filed in federal court within 90 days after receiving it.

​

February 21, 2023

Mobley sued Workday under Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 1981, the Age Discrimination in Employment Act (ADEA), and the Americans with Disabilities Act (ADA), as amended.

​

January 19, 2024

The court dismissed the complaint and allowed Mobley to amend it.

​

April 9, 2024

The EEOC filed an amicus brief arguing that Mobley had plausibly alleged each of his theories for why the federal statutes cover Workday.

​

July 12, 2024

The court allowed the disparate impact claims to proceed on the theory that Workday acted as employers' agent, and it dismissed both the claims that treated Workday as an employment agency and the claims of intentional discrimination. It also dismissed a California law claim with leave to amend.

​

May 16, 2025

The court preliminarily certified a nationwide ADEA collective of applicants aged 40 and over who, from September 24, 2020, applied for jobs through Workday's application platform and "were denied employment recommendations."

​

December 5, 2025

The court denied Workday's motion to stay the notice process for the age collective. The court later noted that the notice produced thousands of opt-ins.

​

January 7, 2026

With the court's permission, three more plaintiffs joined the case, adding sex discrimination claims under Title VII and race, sex, and age discrimination claims under California's Fair Employment and Housing Act (FEHA).

​

March 6, 2026

The court again held that the ADEA's disparate impact provision covers job applicants.

​

June 22, 2026

The court let the FEHA claims and Hughes's disability claim proceed, and it dismissed Rowe's race claim and a claim based on Workday's own hiring.

​

July 2, 2026

The court denied Workday's request to certify for immediate appeal the question whether job applicants may bring ADEA disparate impact claims.

​

July 10 and 15, 2026

The plaintiffs filed the operative Corrected Fourth Amended Complaint, and Workday answered it.

​

September 14, 2026

The plaintiffs moved to certify a class with four subclasses.

​

​

​

What Each Side Says

​

The plaintiffs' allegations

​

The plaintiffs allege that Workday screens applicants with automated systems, trained on historical hiring data, "in lieu of individualized human review." They point to the speed, uniformity, and consistency of their rejections as signs of automated decision-making. Hughes alleges that many rejections arrived within minutes or hours, including one at 2:40 a.m. on a Sunday, and Rowe alleges a rejection at the moment she submitted an application.

​

Their central theory is that Workday acts as an agent of the employers that delegate hiring functions to it. The claims now being litigated are disparate impact claims, which challenge specific practices said to have a significant disproportionate effect on a protected group. They cover race and sex under Title VII, disability under the ADA, age under the ADEA, and race, sex, and age under California's FEHA. The complaint lists the previously dismissed intentional discrimination claims under Title VII, the ADEA, and § 1981, along with a dismissed California aiding-and-abetting claim, by heading only, without repleading them, to preserve them for appeal. It seeks class certification, a declaratory judgment, an injunction, back pay, and damages.

​

Workday's position

​

Workday's answer denies that the federal statutes the plaintiffs invoke cover it. It denies that its "automated screening tools . . . disproportionately exclude" the groups named in the plaintiffs' allegations. To Hughes's and Rowe's allegations about the timing of their rejections, it responds that it does not have enough information to admit or deny them and therefore denies them.

​

In earlier briefing, Workday stated that customers "alone choose to use or ignore the AI functionality" and that Workday "does not recommend, screen out, or otherwise assess or predict applicants' likelihood of success in a role." Its answer pleads, as affirmative defenses, that any challenged action was "job-related and consistent with business necessity" and, for the age claims, that any challenged practice was based on "reasonable factors other than age." Workday has also argued that the ADEA's disparate impact provision does not protect job applicants, and it asked the court to certify that question for immediate appeal.

​

​

The Evidence Filed So Far

​

With their class certification motion, the plaintiffs filed three expert reports: one on how machine-learning screening tools are built, trained, and checked for bias; one presenting statistical analyses; and one proposing a method to estimate class-wide damages if disparate impact is found. Workday's opposition is due November 10, 2026.

​

The plaintiffs' statistical analysis

​

The plaintiffs' statistical and test validation expert, Dr. Charles A. Scherbaum, is a professor of industrial and organizational psychology. He analyzed data produced by Workday on HiredScore Spotlight grades and on outcomes in "Workday's hiring process," which he considers a reasonable data source because the Spotlight algorithm is standardized for all users. He reports that at one stage of that process, the difference in rejection rates between applications from Black and White applicants measured 18.356 standard deviations, and the difference between applications from women and men measured 15.306 standard deviations. He states that a two-tailed result of 1.96 standard deviations or more is generally treated as statistically significant. The public version of his report redacts much of the underlying information, including his summary of opinions and the application counts, rejection counts, and rejection rates in the tables reporting those results.

​

The published bias audits

​

Published summaries of three audits of HiredScore's grading system, completed in July 2023, February 2024, and September 2024, each state that "[t]here is no evidence of disparate impact based on the calculated impact ratios." A 2025 audit commissioned by TIAA found no impact ratios below 0.80, and a Pfizer audit summary reports ratios of 80 percent or higher for every group it analyzed. The plaintiffs cite the three published audits, each of which used a random sample of ten large customers, and the TIAA audit, which they describe as resting on a cross-employer sample that Workday assembled, to argue that the tools' effects can be measured with data pooled across employers.

​

Two different measures

​

Under the federal Uniform Guidelines on Employee Selection Procedures, federal enforcement agencies generally regard a group's selection rate below four-fifths (80 percent) of the highest group's rate as evidence of adverse impact, and smaller differences may also count where they are "significant in both statistical and practical terms." The published audits measure impact ratios against that 80 percent threshold, while Dr. Scherbaum reports tests of statistical significance.

​

​

What the Court Has and Has Not Decided

​

The court's main rulings so far have addressed whether the claims were adequately pleaded and whether they may proceed on a group basis. The case is in discovery, and class certification is being briefed.

​

Motions to dismiss. At this stage a court accepts the complaint's factual allegations as true and asks whether they state a plausible claim. The court reasoned that "[d]rawing an artificial distinction between software decisionmakers and human decisionmakers would potentially gut anti-discrimination laws in the modern era." It also observed that many software vendors would not qualify as agents, giving the example of a spreadsheet program that an employer uses to sort applicants by birthdate and screen out those over 40; in that example, the court said, the spreadsheet takes no part in deciding whom to hire. Its observation that rejection emails sent in the middle of the night gave rise to "a plausible inference that the decision was automated" was made under the pleading standard.

​

Preliminary certification. The court described the standard at this step as "loosely akin to a plausibility standard" and explained, quoting appellate precedent, that the "sole consequence" of preliminary certification is court-approved notice to people who may want to join the case. After additional discovery, Workday may move to decertify the collective.

​

Class certification (pending). Under Supreme Court precedent, a court deciding class certification considers merits questions only to the extent they bear on whether the requirements of Federal Rule of Civil Procedure 23 are met.

​

Questions still contested. Because these rulings applied the pleading and preliminary certification standards, they did not resolve contested facts, including whether Workday's tools caused any disparity, whether Workday in fact acted as its customers' agent, whether any challenged practice is justified by business necessity, and whether anyone is owed damages. Workday's answer disputes each of these points.

​

​

What the Case Means Under Federal Law

​

The case concerns how three federal statutes apply when employers use a software vendor's tools in hiring. Title VII bars employment discrimination because of race, color, religion, sex, or national origin; the ADEA's protections are limited to individuals at least 40 years old; and Title I of the ADA bars discrimination against qualified individuals on the basis of disability. Each statute defines "employer" to include "any agent" of an employer.

​

Under Title VII, a disparate impact plaintiff must show that a specific employment practice produces a disparate impact, and the employer may then defend the practice as "job related for the position in question and consistent with business necessity." A plaintiff may also prevail by making the required showing about an alternative employment practice that the employer refuses to adopt. The ADA bars selection criteria that "screen out or tend to screen out" people with disabilities unless the criteria are shown to be "job-related for the position in question" and "consistent with business necessity." The ADEA bars practices that adversely affect a person "because of such individual's age" but permits actions "based on reasonable factors other than age."

​

The district court has resolved two central legal questions at the pleading stage, and it declined to certify one of them for immediate appeal:

​

  1. Vendor liability. Can a software provider be liable as an employer's agent when employers delegate hiring functions to its tools? The court held that the allegations against Workday plausibly stated such liability.
     

  2. Applicants under the ADEA. Does the ADEA's disparate impact provision, which refers to an employer's "employees," protect job applicants? The court held that it does. In declining to certify the question for immediate appeal, it acknowledged "substantial ground for difference of opinion," comparing a 2019 en banc decision of the Seventh Circuit with the district court decision it had followed.

​

Federal executive branch positions have changed during the case. The EEOC supported Mobley's coverage theories in April 2024. In April 2025, Executive Order 14281 declared a policy "to eliminate the use of disparate-impact liability in all contexts to the maximum degree possible" and directed federal agencies to "deprioritize enforcement of all statutes and regulations to the extent they include disparate-impact liability," naming 42 U.S.C. § 2000e-2 among them. The order is addressed to federal agencies and states that it creates no legally enforceable rights. Mobley is a suit brought by private plaintiffs.

​

​

New Jersey: Rights Under the Law Against Discrimination

​

Mobley is a federal case in California that asserts federal and California claims, not claims under New Jersey law. New Jersey's rules are summarized here because Agard Research Associates Inc. is based in New Jersey.

​

The New Jersey Law Against Discrimination (LAD) bars employers from refusing to hire because of protected characteristics, including race, national origin, age, sex, gender identity or expression, disability, and "liability for service in the Armed Forces of the United States." It also bars any person, "whether an employer or an employee or not," from aiding or abetting a prohibited act.

​

Guidance issued in January 2025 by the New Jersey Office of the Attorney General and the Division on Civil Rights (DCR) states that the LAD "applies to algorithmic discrimination in the same way it has long applied to other discriminatory conduct." It adds that a covered entity may be liable for algorithmic discrimination "even if the covered entity uses a tool it did not develop." Under the guidance, a practice that results in a disparate impact is prohibited unless it is "necessary to achieve a substantial, legitimate, nondiscriminatory interest" and no less discriminatory alternative would serve that interest. In assessing whether a less discriminatory alternative exists, the guidance treats an entity's bias testing, or its review of alternatives, as potentially relevant evidence. The guidance states that it "does not impose any new or additional requirements that are not included in the LAD."

​

DCR's rules on disparate impact discrimination, N.J.A.C. 13:16, took effect on December 15, 2025. Among other things, they:

​

  • define "automated employment decision tools" broadly, including software that produces scores, rankings, classifications, or recommendations that employers use in hiring and other employment decisions;
     

  • place the initial burden on the complainant to show a disparate impact, then on the employer to show that the practice is "necessary to achieve a substantial, legitimate, nondiscriminatory interest," and then on the complainant to show a less discriminatory alternative, with each step requiring empirical evidence;
     

  • require an employer whose practice results in a disparate impact and relies on a vendor's products or systems to "take reasonable steps" to ensure that the vendor's conduct and systems comply with the LAD and the rules;
     

  • identify automated employment decision tools generally, tools that screen applicants by schedule, and tools not adequately tested for adverse effects on protected groups as practices that may have a disparate impact; and
     

  • apply the federal Uniform Guidelines on Employee Selection Procedures to every characteristic the LAD protects.

​

A bill that would regulate automated employment decision tools, A2726, was introduced and referred to the Assembly Science, Innovation and Technology Committee on January 13, 2026. As of October 26, 2026, the Legislature's bill history lists no later action.

​

​

What Official Sources Say to New Jersey Employers

​

The points below restate official New Jersey and federal sources. They are not legal advice.

​

  1. Evaluate tools before and after use. DCR's guidance states that it is critical for employers, and for the developers and vendors of these tools, to "carefully consider and evaluate the design and testing" of automated tools before deploying them and to keep evaluating them after deployment.
     

  2. Test for bias and consider alternatives. According to the guidance, an employer's bias testing and its review of alternatives can be relevant evidence on whether a less discriminatory alternative exists.
     

  3. Oversee vendors. When an employer's practice results in a disparate impact and relies on a vendor's products or systems, DCR's rules require the employer to take reasonable steps to ensure those systems comply, and the guidance states that an employer may be liable even for a tool that it did not build itself.
     

  4. Test before deployment. DCR's rules list an employer's use of a tool that was not adequately tested beforehand for adverse effects on a protected class as a practice that may have a disparate impact.
     

  5. Provide a way to request accommodations. A tool that screens applicants by schedule "must include a mechanism for applicants to request a reasonable accommodation," and both DCR's guidance and the ADA address reasonable accommodation more broadly.
     

  6. Keep impact records. The Uniform Guidelines call for employers to keep records showing the impact of their selection procedures by race, sex, and ethnic group, and DCR's rules apply those guidelines to every characteristic the LAD protects.

​

​

What Comes Next

​

Workday's opposition to class certification is due November 10, 2026. The plaintiffs' notice of motion lists a hearing date of March 9, 2027. Discovery is continuing: in October 2026, the magistrate judge allowed Workday to depose fifteen of the plaintiffs' thirty-eight declarants and noted that the age collective has more than 15,000 opt-in members. A ruling on class certification will decide whether the claims may proceed on a class basis, not whether Workday is liable.

​​

​

A fully cited version of this post, with footnotes and a bibliography, is available on this page. Court filings link to the public docket on CourtListener.

About This Report

Agard Research Associates Inc. (ARA) is a nonpartisan, nonprofit research institute in New Jersey. This report describes allegations, court filings, and rulings in Mobley v. Workday, Inc. as of October 26, 2026, and its footnotes and bibliography give complete citations for every statement. It neither supports nor opposes any party, candidate, or legislation, and it is general information, not legal advice. Version 1.0, October 2026.

Copyright Information

Mobley v. Workday: What the AI Hiring Lawsuit Alleges and What the Court Has Decided © 2026 by Agard Research Associates Inc. is licensed under CC BY-NC-ND 4.0.

​

To view a copy of this license, visit https://creativecommons.org/licenses/by-nc-nd/4.0/

Reasearch Organization Registry (ROR) Logo

859 US Highway 130 

Ste 5 PMB 5020 

East Windsor, NJ 08520-2900

© 2026 Agard Research Associates Inc. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. 


© 2026 Agard Research Associates Inc. All written, visual, and digital content on this site is protected by U.S. and international copyright law. Unauthorized reproduction or distribution is prohibited.​​

bottom of page