Mobley v. Workday: when an ATS vendor becomes an employer’s agent
The Mobley v. Workday, Inc. lawsuit now defines the risk baseline for every HR technology decision maker using AI-driven applicant tracking systems and automated screening tools. In Mobley v. Workday, Inc., Judge Rita Lin of the Northern District of California held at the motion-to-dismiss stage that Workday can be treated as an employer’s “agent” under California’s Fair Employment and Housing Act (FEHA) because its artificial intelligence tools actively participate in the hiring process by scoring, ranking, and recommending candidates. As the court explained, Workday’s tools allegedly “perform functions that are traditionally carried out by employers,” meaning the platform is no longer a neutral conduit but a co-architect of hiring decisions. That single theory of liability turns Workday and similar AI hiring vendors into potential co-defendants whenever job applicants allege discrimination or disparate impact in algorithmic decision making.
The First Amended Complaint in Mobley v. Workday, Inc., No. 3:23-cv-00770-RSL (N.D. Cal.), alleges that Workday’s applicant screening tools produced discrimination against older, Black, and disabled candidates across multiple employers and job families. For example, Mobley pleads that he applied to more than 80 positions between 2018 and 2023 and was rejected at the initial screening stage despite meeting posted qualifications, while non-Black and non-disabled applicants allegedly advanced at materially higher rates (FAC ¶¶ 34–47). Judge Lin’s March 25, 2024 order (ECF No. 63) emphasizes that these claims are plausible at the pleading stage because the same screening system allegedly affected outcomes across different organizations and roles, noting that Mobley “plausibly alleges that Workday’s tools screened him out for positions at multiple employers” (Order at 9–10). Because Workday designs and operates its screening tools from California, the court held that out-of-state employers can still face FEHA exposure when they rely on this third-party platform for applicant screening and hiring decisions (Order at 13–15). That holding extends AI hiring vendor liability and algorithmic discrimination risk far beyond California-headquartered companies. The motion-to-dismiss ruling also signals that other providers such as Eightfold, HireVue, and Pymetrics may face similar discrimination claims when their tools score, rank, or recommend job applicants in ways that create disparate impact under federal anti-discrimination and state civil rights laws.
For employers, the message is blunt: you cannot outsource compliance, bias risk, or human oversight to a vendor, even a large platform like Workday with sophisticated artificial intelligence capabilities. When an employer configures screening tools, defines job requirements, and accepts automated recommendations, it shares legal liability with vendors for any discrimination claims that follow, including potential Title VII exposure in federal court alongside FEHA claims in California court. The Mobley precedent therefore reframes every AI hiring tool as a joint enterprise in employment decision making, not a simple software purchase, and it forces employers and vendors to align on bias audits, data governance, and defensible hiring process design. In practical terms, any ATS or AI screening engine that shapes who advances, who is rejected, and who receives offers must now be treated as part of the employer’s decision-making apparatus, not an external black box.
From procurement to co liability: how contracts, audits, and data must change
Talent acquisition leaders now need to treat every ATS and AI screening contract as a shared risk instrument, not just a software subscription. The Mobley v. Workday ruling makes it clear that when vendors provide algorithmic decision engines for applicant screening, they become deeply embedded in the hiring process and therefore share responsibility for any discrimination or disparate impact that flows from those tools. Judge Lin’s order notes that Mobley “plausibly alleges that Workday’s tools are a but-for and proximate cause of the alleged discrimination” (Order at 11), which underscores why procurement and legal teams must assume that AI hiring platforms can be named alongside employers in future complaints and build that assumption into every master services agreement and data processing addendum.
Four operational moves stand out for HR technology decision makers who want to stay ahead of court and federal enforcement trends:
1. Rebuild contracts around shared liability. Audit every contract with Workday and other AI recruitment vendors to ensure that algorithmic discrimination, Title VII exposure, and FEHA-style theories of liability are explicitly addressed, including clear duties for both employer and vendor when screening tools are updated or re-trained. Contracts should spell out indemnification for discrimination claims, notification timelines when models change, and obligations to cooperate in responding to EEOC charges or state civil rights investigations.
2. Demand transparent model documentation. Require model cards, bias testing summaries, and documentation of algorithmic decision logic from vendors, and pair those artefacts with independent bias audits under attorney supervision. The court’s order highlights that legal advice around testing and mitigation can fall within attorney–client privilege and work-product protection when counsel directs the analysis for the purpose of giving legal guidance (Order at 18–19), which means employers should structure audits through counsel to preserve that protection where possible.
3. Secure data and audit rights for long-term analysis. Secure contractual rights to conduct or commission third-party audits of applicant screening tools, including the right to inspect training data, feature importance, and rejection reasons for job applicants across at least six years of hiring data to detect disparate impact patterns. For example, employers should be able to compare pass-through rates by age, race, disability status, and other protected characteristics over time to identify whether an AI screening model is systematically excluding certain groups. A basic benchmark is to test whether selection rates for any protected group fall below 80% of the highest group’s rate, mirroring the “four-fifths rule” often referenced in disparate impact analysis.
4. Align internal teams around a unified oversight framework. Align internal compliance, legal, and talent acquisition équipes around a unified hiring process framework that embeds human oversight at every high-stakes decision point, supported by structured screening interviews and documented criteria, as outlined in this playbook on how a structured screening interview shapes smarter hiring decisions. In parallel, HRIS leaders should benchmark their governance model against emerging regulatory frameworks such as the EU AI Act, using resources like this analysis of the EU AI Act compliance countdown for recruiting to anticipate how cross-border enforcement will interact with AI hiring vendor liability and discrimination dynamics.
Redesigning AI enabled hiring: governance, human oversight, and future proof stacks
Mobley v. Workday is not just a litigation story; it is a blueprint for how AI-enabled hiring must be governed. The decision tells employers that every algorithmic decision about job applicants must be traceable, explainable, and anchored in validated job-related criteria, with humans retaining final accountability for hiring decisions rather than rubber-stamping scores from opaque tools. Judge Lin’s order underscores that when a vendor’s system screens, ranks, and recommends candidates, it can be treated as acting “on behalf of” the employer (Order at 7–9), which raises the stakes for how those systems are designed and monitored. That means HR technology leaders need a decision framework for when to buy, build, or pause artificial intelligence capabilities in their hiring stack, such as the one outlined in this guide to AI in recruitment decision frameworks.
Practically, employers should map every step of their hiring process, from sourcing to final offer, and label where vendors, algorithms, and humans each influence outcomes for job applicants. At each step where screening tools or algorithmic decision engines operate, they should define explicit human oversight controls, such as mandatory review of automated rejections, periodic bias audits of pass-through rates by protected class, and documented rationales when recruiters override or accept AI recommendations in high-volume hiring. A simple checklist can help: identify each automated decision point, assign a human owner, specify review cadence, and log every material model change with its expected impact on selection rates. Those controls should be aligned with federal anti-discrimination standards, Title VII guidance, and state-level discrimination laws, so that if an amended complaint or motion to dismiss arises, the employer can show the court a coherent compliance architecture rather than ad hoc fixes. A defensible record of how AI tools are evaluated, tuned, and supervised will matter as much as the tools themselves.
Finally, HR leaders should treat vendors like Workday, Eightfold, and HireVue as strategic partners in risk management, not just providers of convenient tools for faster hiring. That means pressing vendors to share detailed information about their screening tools, bias testing methodologies, and liability assumptions, while also insisting on joint governance forums where employer and vendor review discrimination claims data, disparate impact metrics, and remediation plans across the entire employment lifecycle. In this new environment, the strongest talent acquisition strategies will be built on transparent AI, disciplined human oversight, and contracts that recognize a simple truth: when your ATS shapes who gets hired, your vendor is no longer just a platform, it is part of your decision-making brain.