Inside the Mobley v. Workday Rulings: What 2026’s AI Hiring Bias Case Means for HR Teams

Roughly 1.1 billion rejected job applications appear in the record surrounding Mobley v. Workday, not as 1.1 billion proven instances of discrimination, and not necessarily as 1.1 billion unique applicants, but as the number Workday represented had been rejected using relevant software tools during the period at issue. That scale helps explain why this federal AI hiring discrimination case has become unusually consequential for HR, procurement, employment counsel, and vendors that automate parts of applicant screening.

By mid-2026, Mobley v. Workday, Inc., Case No. 3:23-cv-00770-RFL, before U.S. District Judge Rita F. Lin in the Northern District of California, had survived multiple efforts to narrow or dismiss its central discrimination theories. The plaintiffs allege that Workday's algorithm-based applicant-screening systems discriminated on the basis of race, age, and disability; later amended pleadings expanded the group of named plaintiffs and claims.

For an HR executive team, however, the useful question is not whether the words “AI lawsuit” sound alarming. It is what the court has actually decided, what it has not decided, and which assumptions in your vendor-governance model no longer look safe.

There is also an important timeline correction to make before analyzing the Workday AI hiring lawsuit 2026 developments. The nationwide ADEA collective was conditionally certified on May 16, 2025, not March 6, 2026; February 2026 brought court-authorized notice to potential collective members, while the March 6 order addressed dismissal issues including whether job applicants may pursue ADEA disparate-impact claims.

The June timeline also needs precision. June 16, 2026 was a hearing, not the ruling: Reuters reported that Judge Lin appeared skeptical of Workday's position on California law but had not yet ruled; the consequential order largely denying Workday's motion followed on June 22, 2026.

That distinction matters. HR compliance depends on reading procedural history accurately: “the claim survives” is not the same statement as “the defendant discriminated.”

This article therefore focuses on this specific case rather than repeating a general AI-recruiting-tool taxonomy or another overview of AI regulation. It explains what Mobley alleges, the sequence of rulings that produced the case's current posture, what the “AI vendor as agent” theory means, and what you should reconsider in vendor contracts and governance without assuming an outcome the court has not reached.

Publication-status warning: Mobley v. Workday remains active litigation. The discussion below reflects rulings and docket developments located through August 14, 2026, including a July 1 order directing the filing of a Fourth Amended Complaint after the June ruling; no final judgment on whether Workday actually discriminated is treated here as having occurred. Verify the current docket through PACER, another reliable legal database, or the Northern District of California immediately before publication because later motions, settlement developments, certification decisions, or appellate activity can supersede this analysis.

This is HR-compliance education, not legal advice for a particular employer, contract, applicant population, or jurisdiction.

The Case That's Turning AI Hiring Vendors Into Legal Defendants

The core allegation in Mobley v. Workday is unusually direct. Derek Mobley filed suit in 2023 alleging that Workday's algorithm-based applicant-screening tools discriminated against him and similarly situated applicants based on protected characteristics; the March 6, 2026 court order describes the surviving federal disparate-impact claims as race under Title VII, disability under the ADA, and age under the ADEA.

Mobley alleges that he applied to more than 100 jobs through employers using Workday and received no offers. Legal reporting describes him as Black, over 40, and having anxiety and depression, and notes that the litigation later added three named plaintiffs and additional discrimination theories.

What makes the case different from an ordinary hiring-discrimination dispute is who Mobley sued. Rather than building the principal theory around each prospective employer that rejected him, the plaintiffs have sought to hold Workday itself responsible for allegedly discriminatory screening activity performed through its technology.

That choice puts AI vendor liability in employment discrimination at the center of the dispute. It asks whether a technology company can become legally relevant under statutes normally associated with employer hiring decisions when employers delegate enough of the screening, ranking, scoring, recommendation, or rejection process to its systems.

For HR teams, this is why Mobley deserves separate treatment from the broader landscape of AI recruiting tools already covered on this site. Refonte Learning's August 8, 2026 article maps sourcing, screening, conversational tools, interview intelligence, and ATS-related AI; the present issue is narrower and more legal: what happens when the software provider itself becomes a discrimination defendant because of the role its system allegedly plays in selection?

The litigation's scale makes that question harder to dismiss as a one-applicant edge case. In connection with the ADEA collective proceedings, Workday represented that 1.1 billion applications were rejected using its software tools during the relevant period, and Proskauer reported that Workday argued the potential group could therefore extend into the hundreds of millions. That figure describes applications processed or rejected in the relevant context; it does not establish that 1.1 billion discriminatory acts occurred.

The procedural history also looks different once the dates are correctly separated:

The table corrects an important misconception in summaries of the Workday age discrimination ruling. March 6 did not create the nationwide collective; May 16, 2025 did. The March decision instead preserved the plaintiffs' ability to pursue disparate-impact age claims as job applicants, and the February 2026 notice process operationalized the previously certified collective.

That difference is not academic. When you brief an executive team, “the judge certified nationwide age claims last week” communicates a materially different litigation event from “the collective had already been conditionally certified, and the court has now rejected another statutory defense.”

The latter is what the record supports.

The 2026 Rulings, in Order

You can understand the HR compliance AI hiring tools 2026 implications more clearly by treating Mobley as a sequence of litigation gates rather than one giant “AI discrimination ruling.” Each gate has a different legal function and therefore a different compliance lesson.

The age-collective development. The court conditionally certified the ADEA collective on May 16, 2025 after finding, for purposes of that procedural stage, sufficient commonality around allegations concerning Workday's AI recommendation system and its use to score, sort, rank, or screen applicants. Proskauer reports that Workday's objections based on differences among client configurations and applicants did not defeat conditional certification at that stage.

The practical effect of an ADEA collective mechanism is that similarly situated individuals can opt into the federal age claim rather than leaving the dispute confined to Derek Mobley. Plaintiffs' counsel stated on February 17, 2026 that the authorized notice covered people who applied through Workday's platform from September 24, 2020 onward and were at least 40 at the time, with March 7, 2026 set as the stated response deadline.

For HR, the scale lesson is not “every rejection is now a claim.” It is that centralized automated selection practices can create a common factual mechanism around which plaintiffs attempt to aggregate large numbers of applicant experiences, making data retention, model history, configuration records, and validation evidence more consequential than they appear when you look at one hiring requisition at a time. That is a compliance inference from the collective proceedings, not a holding that any particular employer's records are deficient.

The March 6 order. Judge Lin's March decision expressly considered Workday's contention that job applicants cannot bring disparate-impact claims under the ADEA and rejected it, reiterating the court's earlier position. The order therefore denied Workday's motion to dismiss the ADEA claim on that basis.

The same order initially went the other direction on California's Fair Employment and Housing Act. Judge Lin held that the Second Amended Complaint did not contain enough nonconclusory factual allegations connecting the alleged unlawful conduct to California, so she dismissed those FEHA claims with leave to amend; notably, the order explained that facts described in plaintiffs' briefing, if properly alleged, could be sufficient to establish the required California nexus.

That procedural detail sets up June.

The June 16 hearing. Reuters reported that Workday argued California law should not govern screenings involving applicants outside California applying to employers and jobs outside the state. Judge Lin questioned that position by focusing on where the allegedly unlawful conduct itself occurred, but Reuters explicitly reported that she did not say at the hearing when she would rule.

That means an article stating “the judge ruled on June 16” would turn judicial questioning into a holding. For a compliance audience, that is precisely the kind of legal-news shortcut to avoid.

The June 22 order. Six days later, Judge Lin largely denied Workday's effort to dismiss the recently amended claims. Reuters reports that she rejected Workday's argument that California anti-discrimination law categorically could not apply when the screened applicants and jobs were outside California, because plaintiffs alleged that Workday itself participated in the challenged conduct from its California headquarters.

The important nuance is not “California law now applies to every out-of-state employer that buys California software.” The ruling is narrower: at the pleading stage, the plaintiffs alleged enough of a nexus between California and Workday's own allegedly unlawful conduct to keep the FEHA theories alive even though affected applicants could be outside California and applying for jobs elsewhere.

The court also allowed an ADA theory alleging that the software could screen through proxy indicators associated with disability or illness to continue, while dismissing an Asian-American discrimination claim because of the procedure used to add it. Workday publicly denied the allegations, saying its recruiting tools assess job qualifications, do not make hiring decisions, and undergo responsible-AI testing.

On July 1, the court issued a follow-up order implementing the June decision, adopted modified strikes, and directed plaintiffs to file a clean Fourth Amended Complaint; Workday's answer was then due July 15. That docket activity reinforces the central procedural point: the lawsuit continued; the court had not entered judgment that discrimination occurred.

For an executive briefing, I would reduce the entire sequence to four statements:

• The nationwide ADEA collective dates to May 16, 2025, with the notice process continuing into early 2026.

• The March 6, 2026 order preserved the applicant disparate-impact theory under the ADEA while initially requiring better pleading for the California-law claims.

• June 16 was the hearing; June 22 was the ruling that allowed key amended California and disability theories to continue.

• None of those events is a final determination that Workday's systems discriminated; Workday disputes the allegations.

• That is less dramatic than “AI vendor found liable,” but it is more useful, and legally accurate.

Why the Agent Theory Matters Beyond Workday

The most consequential part of Mobley for organizations that have never purchased Workday may be the agent theory. The court had already allowed that theory to proceed in 2024, and legal analysts continued emphasizing it in 2026 because it offers a route by which a technology vendor performing delegated hiring functions may come within statutory definitions that reach an employer's agents.

In practical terms, the allegation is not merely “an employer happened to use software while making a decision.” It is that employers allegedly delegated functions traditionally associated with recruiting and selection, such as screening, ranking, scoring, recommending, or rejecting applicants, and the vendor's systems participated sufficiently in those functions for the complaint to plausibly treat Workday as an agent.

That does not mean every ATS, assessment platform, résumé-ranking system, or AI service automatically becomes an employer's statutory agent. The court's determination arose from the allegations before it and at a pleading stage, and the eventual merits record can matter greatly to whether the plaintiffs prove the theory.

It also does not mean the employer vanishes from the legal picture because the vendor can face claims. The more defensible compliance reading is the opposite: vendor-side exposure and employer-side exposure can coexist conceptually, so HR should stop treating “who is liable?” as a binary procurement question with only one possible responsible party. JD Supra's June 5 analysis similarly warns employers that the agent theory undermines the instinct to treat a vendor flaw as automatically external to the employer's hiring process.

Think about the distinction this way:

The key legal point is not that a computer becomes a human recruiter in a literal sense. It is that outsourcing a decision function does not necessarily make that function legally disappear merely because software performs it through a vendor relationship.

SHRM's July 1, 2026 analysis similarly describes Mobley as a warning that organizations cannot simply treat HR AI as a black box managed exclusively by the vendor. SHRM traces the agent issue back to the 2024 ruling and the May 2025 ADEA collective certification, again confirming that the doctrinal shift predates the June 2026 pleading decision.

For HR leadership, that changes the internal ownership map. An automated screening purchase may still travel through procurement and IT, but the associated governance now has to reach employment-law analysis, talent-acquisition process design, accessibility, data/analytics review, record preservation, and contract negotiation where those functions are implicated.

The court did not order companies to create that governance structure. It is a risk-control inference from what happens when a vendor's alleged role in the hiring decision becomes important enough for the vendor itself to remain a discrimination defendant.

What this case does not mean deserves equal emphasis.

Measured compliance means holding both ideas at once: this is not a final liability judgment, and it is still a serious case-law signal. Treating it as settled nationwide law would overstate the rulings; treating it as meaningless because “nothing has been proven yet” would understate why motions to dismiss, collective proceedings, and discovery matter in high-volume discrimination litigation.

That is the right posture for AI hiring bias legal risk in 2026: neither panic nor reassurance unsupported by the record.

The Regulatory Backdrop This Case Is Landing In

Mobley matters partly because it demonstrates something easy to miss when every AI-compliance discussion becomes a list of newly enacted AI statutes: old discrimination laws can do substantial work without waiting for a dedicated “AI hiring law.” The case's federal theories arise under Title VII, the ADA, and the ADEA, while the amended state claims invoke California's FEHA.

That context differs from the general strategy discussion in how HR management is transforming the workforce in 2026. Here, the important question is narrower: what legal protections and enforcement mechanisms remain relevant to an allegedly discriminatory hiring decision when AI-specific policy is shifting?

The federal guidance picture changed in January 2025. K&L Gates reports that the EEOC removed AI-related guidance from its website on January 27, 2025, including May 2023 material discussing how federal anti-discrimination law can apply when employers use AI in hiring and other employment decisions.

Do not translate that event into “federal discrimination law no longer covers AI.” The statutes did not disappear, and an EEOC Strategic Enforcement Plan page still available in 2026 identifies technology, including AI and machine learning used in recruitment, screening, and hiring, as capable of contributing to discrimination against protected groups.

The accurate formulation is therefore: specific EEOC AI technical-assistance material was removed, while underlying employment-discrimination law remains operative and litigation continues to test its application to automated systems. Mobley is an unusually concrete example because the court is addressing statutory theories in an actual adversarial case, not issuing general-purpose vendor guidance.

The European timeline also changed substantially during 2026. The EU Council announced on June 29 that the new application date for stand-alone high-risk AI systems would be December 2, 2027, rather than August 2, 2026; employment-related systems listed among the AI Act's high-risk categories fall within the relevant stand-alone high-risk framework, while certain product-embedded systems received a later August 2, 2028 date.

The legislative procedure has since been completed: the European Parliament's procedure record identifies Regulation (EU) 2026/1744, published in the Official Journal on July 24, 2026, as the final Digital Omnibus on AI act, and records December 2, 2027 for stand-alone high-risk categories including employment.

For HR teams that had August 2, 2026 circled as the decisive date for Annex III employment systems, that is a real schedule change. It is not a reason to defer your Mobley-related analysis until December 2027: the U.S. lawsuit is moving under existing anti-discrimination law independently of the EU implementation calendar.

That is the most useful regulatory context to retain:

This is also why this article does not rebuild generic explainers for Illinois, Colorado, Connecticut, California, or New York City's Local Law 144. Those regimes matter to a multistate compliance map, but Mobley contributes a different insight: a plaintiff does not necessarily need a statute with “artificial intelligence” in its title to challenge algorithmic screening behavior.

For an executive audience, that is the cleanest answer to “Can we wait until the AI rules settle down?” No. Regulatory timetables can move; the litigation exposure created by existing discrimination statutes can move on a different calendar entirely.

What HR Teams Should Actually Do Differently Now

The best response to Mobley is not to shut down every automated recruiting feature. The rulings do not ban AI hiring tools, and Workday continues to deny that its technology discriminates or makes the hiring decisions alleged by the plaintiffs.

The better response is to upgrade AI hiring from “software procurement” to “employment-decision governance.” SHRM has separately recommended that employers vet vendors closely, ask how algorithms are created, understand audit practices, and examine vendor agreements in light of Mobley.

Start with four controls:

First, ask for evidence, not adjectives. “Bias tested,” “responsible AI,” “compliant,” “explainable,” and “human in the loop” can be useful starting labels, but they do not tell you which population was tested, which protected characteristics were evaluated, which product version was assessed, what customer-specific configuration changes the output, or what happened after the vendor updated the model.

Your due-diligence file should be capable of answering a more disciplined set of questions:

• What exact feature scores, ranks, filters, recommends, suppresses, or automatically rejects an applicant?

• Which data fields and derived signals materially influence that output?

• Which protected-group impact tests does the vendor run, how often, and against what comparison population?

• Can customer configuration materially alter the vendor's validation conclusions?

• What happens when an update changes model behavior, scoring logic, features, thresholds, or training data?

• Can you obtain sufficient documentation to investigate a complaint without depending entirely on the vendor's willingness to cooperate?

• How do accommodation requests or disability-related application patterns interact with the automated workflow?

• What evidence can you preserve to reconstruct the system's role in a decision months or years later?

Those questions follow from the type of delegated selection functions and discovery issues that Mobley has brought into litigation, but they are risk-management recommendations rather than court-ordered requirements. JD Supra reports that plaintiffs have sought algorithmic training, testing, methodology, code, and related records in discovery, illustrating why reconstructability becomes significant when the model itself is contested.

Second, stop reading indemnification as a synonym for “no liability.” An indemnity is a contractual allocation mechanism; it does not itself determine whether a discrimination statute applies to your organization or the vendor.

JD Supra's June 2026 defense-side analysis specifically flags indemnification caps as a concern in AI subscription contracts, arguing that contractual caps designed around subscription fees may bear little relationship to the exposure alleged in a large collective case. That is one analyst's contract-risk assessment, not a universal statement about every vendor agreement, but it identifies the provisions HR and counsel should actually read.

At minimum, contract review should identify:

Third, preserve an employer-side audit trail. Do not assume the vendor's logs will always contain everything your organization needs or remain available for the entire limitations and litigation period relevant to a later claim.

A useful internal record can identify the requisition, screening feature, vendor/model version, configuration and threshold, criteria approved by the employer, dates used, downstream human review, override activity, documented accommodation handling, testing results, and material vendor changes. The precise retention period and content should come from applicable law and counsel because jurisdiction, claim type, privacy obligations, and litigation holds can differ.

Fourth, define accountability before deployment. Someone inside the organization should know who can suspend the automated feature, who reviews impact results, who approves configuration changes, who receives complaints, who contacts the vendor, and who determines whether employment counsel must become involved.

This is where the trends, challenges, and future-proof strategies shaping HR in 2026 become operational rather than abstract. Technology fluency has limited value if nobody can answer who owns the consequences of a selection system after procurement signs the contract.

Most importantly, do not let the vendor-agent theory produce a new version of the same mistake. The correct takeaway is not “the vendor can be liable, therefore we are safe”; it is “the vendor may no longer be legally invisible, therefore both sides need clearer evidence, responsibilities, and contractual allocation.”

That is a materially different governance posture.

Skills, Certifications, and Common Mistakes for AI-Vendor Legal Risk

The human resource management skills 2026 employers need around AI do not require HR professionals to become litigators. They do require enough legal literacy to know the difference between an allegation, a pleading-stage ruling, collective certification, a merits judgment, a statutory theory, and a risk-management recommendation.

That distinction ranks first because Mobley is already a case study in how quickly legal shorthand can distort reality. Calling the March 6 order “nationwide certification” or the June 16 hearing “the ruling” produces a compliance memo that is wrong before it reaches the executive team.

A practical priority order looks like this:

These skills fit alongside, rather than replace, the capabilities discussed in talent acquisition trends, challenges, and strategies for the future. Mobley adds a compliance layer: recruiting professionals need to understand not only what a workflow does, but which legally significant hiring function the organization may have delegated to the workflow.

Certification is not a substitute for docket literacy. A case that is still generating amended complaints and motion rulings in 2026 can change faster than a conventional certification curriculum or exam cycle.

SHRM itself is publishing current articles on Mobley, including March and July 2026 coverage, which is a useful illustration of why continuing legal awareness matters alongside formal credentials. The case's developing procedural history is current-event knowledge, not a static rule you should expect every general HR examination to capture by name.

For a portfolio or interview, a stronger signal of this specific competency is your ability to produce a one-page case briefing that cleanly separates facts alleged, issues, rulings, unresolved questions, and HR actions. That exercise shows whether you can translate legal developments into operational decisions without overstating the law.

For example:

Two recurring mistakes deserve special attention.

Mistake: assuming vendor contracts fully transfer discrimination exposure. Contracts can allocate financial and defense obligations between commercial parties, but they do not rewrite statutory definitions or guarantee that only one participant in an employment process can face a claim.

The Mobley agent theory makes that limitation particularly visible. The court allowed claims to proceed based on allegations that Workday performed traditional hiring functions as an employer agent, while JD Supra's analysis cautions employers against assuming vendor-side involvement removes employer-side risk.

The fix is not to demand unlimited indemnification as a reflex. It is to have counsel match the contractual language to the actual selection function, likely causes of action, audit evidence, defense-cost structure, applicable limits, and control of the response.

Mistake: waiting for a single definitive federal AI-hiring manual. The EEOC removed specific AI-related web guidance in January 2025, but existing discrimination statutes remain the legal substrate on which Mobley is proceeding, and the EEOC's still-published Strategic Enforcement Plan discusses discriminatory effects associated with AI and automated screening technology.

The fix is to stop treating “guidance” and “law” as interchangeable. A guidance document can change or disappear; a statutory discrimination claim can continue to be litigated.

A third mistake sits underneath both: outsourcing understanding to the vendor. Your software salesperson, implementation consultant, and customer-success manager can explain the product, but they should not become your only source for determining what your employment-law obligations are.

A fourth mistake is the opposite: overreacting to headlines and disabling a useful system without understanding the ruling. Mobley has not established that all automated screening causes unlawful disparate impact, and the court has not entered a final liability judgment against Workday in the developments reviewed here.

The HR professional who can say, “This raises our diligence threshold, but it does not establish liability,” is more useful to leadership than either the person saying “AI hiring is illegal now” or the person saying “the vendor contract takes care of it.”

Self-Study vs. Structured HR Management: The Refonte Learning Program

You can learn the Mobley docket by reading the orders, reputable legal reporting, and ongoing commentary. What takes longer to build is the underlying employment-law and HR-process literacy that tells you why an ADEA applicant issue differs from an ADA proxy allegation, why an agency theory matters, why a dismissal with leave to amend is not a loss on the merits, and why contract indemnity does not define statutory coverage.

That is where the comparison between self-study and structured HR education becomes relevant.

The structured option should not be sold on a false premise. The Refonte Learning Human Resource Management Program does not say that it teaches Mobley v. Workday, Workday-specific litigation, or AI-vendor-liability doctrine by name; the public curriculum lists broader HR modules and competencies instead.

That is the honest connection. The program's explicitly named Labor Laws & Compliance competency builds the foundational literacy that makes a case like Mobley intelligible, while Talent Acquisition & Recruitment and HR Technology & HRIS Tools provide the process and technology context in which the legal issue arises.

The program runs for 3 months with a published dedication of 8–10 hours per week. Its page recommends a basic understanding of business concepts and lists working toward a bachelor's degree or a higher-level degree as an obligatory admission prerequisite.

Its three educational modules are:

• Introduction to Human Resource Management: HRM fundamentals, organizational role, and key success factors.

• Talent Acquisition and Recruitment Strategies: sourcing, attracting, and selecting talent.

• Performance Management and Employee Engagement: appraisal systems and employee motivation.

The program page lists ten competencies: Talent Acquisition & Recruitment; Employee Onboarding & Engagement; Performance Management & Appraisals; Compensation & Benefits Administration; Workplace Diversity & Inclusion; Labor Laws & Compliance; HR Analytics & Metrics; Conflict Resolution & Employee Relations; Organizational Development; and HR Technology & HRIS Tools.

That mix matters to this article because Mobley sits at an intersection. You need employment-law literacy to understand the statutes, talent-acquisition literacy to understand which function the vendor performed, diversity and inclusion literacy to understand protected-group effects, and technology literacy to ask technically meaningful questions about the system.

Professor Kevin Harris, Senior Advisor at Refonte Learning, is presented by the program page as an HR professional with more than 15 years of experience in talent management and organizational development.

Upon successful completion, Refonte Learning states that students receive a Training Certificate and Certificate of Internship. Students with outstanding performance may also receive a Letter of Recommendation and Certificate of Appreciation, while top performers may receive prizes.

Published career outcomes include HR Manager, Talent Acquisition Specialist, Employee Relations Manager, HR Business Partner, and Training & Development Manager. For readers evaluating where those positions differ in practice, Refonte Learning also covers the difference between an HR generalist and an HR business partner.

The published enrollment fee is $300 as a one-time payment. The page also shows an installment structure of $204 plus $98, and the supplied program information identifies the one-time price as 30% below a $387 list price.

The strongest reason to consider structured HR learning here is not that Mobley itself will remain the dominant AI hiring case forever. Litigation changes.

What persists is the skill of opening the next court order and correctly asking: What statute is involved? Who made or influenced the employment decision? What did the judge actually decide? What remains alleged? What does our own hiring process look like? What evidence do we have? What requires counsel?

That is a more durable definition of human resource management skills in 2026 than memorizing one lawsuit headline.

For a structured foundation in labor-law compliance, talent acquisition, and HR technology, review the Refonte Learning Human Resource Management Program.

FAQ: People Also Ask

Because this litigation remains active, verify all procedural dates against the current federal docket immediately before publication.

What is Mobley v. Workday about?

Mobley v. Workday, Inc., Case No. 3:23-cv-00770-RFL in the U.S. District Court for the Northern District of California, is a federal lawsuit alleging that Workday's algorithm-based applicant-screening tools discriminated against job applicants on protected grounds including age, race, and disability. The March 6, 2026 order identifies surviving federal disparate-impact theories under Title VII, the ADA, and the ADEA, while later amended pleadings added further plaintiffs and state-law allegations.

The frequently cited “1.1 billion” figure refers to Workday's representation in litigation that 1.1 billion applications were rejected using its software tools during the relevant period. It should not be described as 1.1 billion proven victims, unique applicants, or adjudicated instances of discrimination.

What happened in the Mobley v. Workday 2026 rulings?

One frequently repeated timeline needs correction. The court conditionally certified the nationwide ADEA collective on May 16, 2025, and notice to potential collective members was authorized for the 2026 opt-in process; that certification did not occur on March 6, 2026.

On March 6, 2026, Judge Rita Lin again rejected Workday's argument that job applicants may not pursue ADEA disparate-impact claims, while dismissing then-pleaded California FEHA claims with leave to amend because the complaint had not sufficiently alleged a California nexus.

On June 16, the court held a hearing on the California issue but did not rule that day. On June 22, Judge Lin largely denied Workday's next motion, allowing key California-law and ADA theories to proceed based on the amended allegations; the court later issued a July 1 order implementing parts of that decision.

Does this mean Workday was found liable for discrimination?

No. The 2024–2026 decisions discussed here address whether claims may proceed, collective-action procedure, pleading sufficiency, statutory interpretation, and related litigation issues; they do not constitute a final judgment that Workday actually discriminated against the plaintiffs or collective members.

Workday denies the allegations. Reuters reported after the June 22 ruling that the company said its recruiting technology looks at job qualifications rather than protected characteristics, does not make the hiring decisions alleged, and undergoes responsible-AI testing.

Can an AI hiring vendor be held legally responsible, not just the employer?

Mobley demonstrates that this can at least be a viable litigation theory under existing employment-discrimination statutes. The court allowed claims to proceed on allegations that Workday acted as an agent of employer-clients by performing traditional hiring functions such as screening, ranking, scoring, recommending, or rejecting applicants.

That does not establish a categorical rule that every AI vendor is an agent or that Workday will ultimately be liable. Agency and statutory coverage can depend on the functions actually delegated, the facts proved, the relevant statute, the jurisdiction, and subsequent rulings.

What should HR teams do in response to this case?

HR teams should treat automated screening as an employment-governance issue rather than a software purchase alone. Practical steps include requesting documented disparate-impact and validation evidence, reviewing discrimination-related indemnity and liability provisions with counsel, preserving employer-side screening and configuration records, and assigning responsibility for monitoring model changes and investigating complaints. SHRM's own discussion of Mobley similarly advises employers to vet vendors and ask how their algorithms and auditing processes operate.

These actions do not imply that an organization must stop using AI. They reflect the narrower lesson from Mobley: when a vendor may be alleged to perform delegated hiring functions, neither “the vendor owns the algorithm” nor “our contract says the vendor complies” should substitute for independent due diligence.

Does the Refonte Learning Human Resource Management Program teach this specific case or AI-vendor liability law?

No claim should be made that the published curriculum teaches Mobley v. Workday, Workday-specific litigation, or AI-vendor-liability doctrine directly. The program page instead lists broader competencies including Labor Laws & Compliance, Talent Acquisition & Recruitment, Workplace Diversity & Inclusion, and HR Technology & HRIS Tools, alongside three HR-management modules.

The defensible connection is foundational: labor-law and recruitment-process literacy helps an HR professional understand why a case involving alleged AI screening discrimination, disparate impact, protected characteristics, and vendor agency matters. The program runs for three months at 8–10 hours per week and awards a Training Certificate and Certificate of Internship upon successful completion.

Conclusion

The most useful conclusions from the Mobley v. Workday litigation are precise rather than dramatic:

• The nationwide age collective was conditionally certified on May 16, 2025, not March 6, 2026. In 2026, notice proceeded, and the March 6 order rejected Workday's renewed argument that ADEA disparate-impact protection does not extend to applicants.

• The California-law development came in the June 22 ruling, not the June 16 hearing. Judge Lin allowed the amended FEHA theories to continue based on allegations that Workday participated in the challenged conduct from California, even though affected applicants and jobs could be outside the state.

• None of these rulings establishes that Workday actually discriminated. They keep important theories and claims alive while the litigation proceeds, and Workday continues to deny the allegations.

• The vendor-as-agent theory is the case's broadest lesson for HR. A vendor that allegedly performs delegated hiring functions may not be legally invisible merely because the employer bought those functions as software, which makes vendor testing, contractual allocation, documentation, and escalation to employment counsel more important.

The EU AI Act's high-risk timetable may now run to December 2, 2027 for stand-alone high-risk systems, and dedicated EEOC AI web guidance was removed in January 2025, but neither development pauses the application of existing employment-discrimination statutes. Mobley shows why HR teams must be able to read live case law alongside evolving AI regulation instead of waiting for one permanent compliance checklist.

For the labor-law, recruitment, and HR-technology foundation that makes cases like this understandable rather than abstract, the Refonte Learning Human Resource Management Program provides a structured starting point.