Workday AI Hiring
When the screening algorithm became the defendant
What They Said
Workday is the HR system of record for roughly half of the Fortune 500. Its applicant tracking and screening tools — including AI-driven candidate scoring and recommendation features — are used to filter millions of job applications every year. The company has positioned itself as a neutral platform: customers configure the screening criteria, Workday provides the infrastructure, and hiring decisions remain with the employer.
When AI bias concerns surfaced industry-wide, Workday’s public messaging emphasized fairness audits, EEOC compliance, and the customer’s responsibility to set non-discriminatory criteria. The implicit contract with enterprise buyers was clear. The vendor builds the rails. The employer owns the risk.
What Actually Happened
In May 2025, US District Judge Rita Lin certified a nationwide collective action against Workday under the Age Discrimination in Employment Act. The lead plaintiff, Derek Mobley, alleged he had been rejected from more than 100 jobs through Workday’s screening system despite being qualified — and that the pattern was consistent with age-based filtering. The court found enough evidence of common algorithmic decision-making to allow the case to proceed as a class spanning all US applicants over 40 who were screened by Workday’s tools.
The certification was the legal earthquake. Until that ruling, the conventional reading of US employment law treated software vendors as tool providers, not as employers or employment agents. The Mobley case pushed the theory that Workday itself was acting as an “agent” of the employers using its system, because its algorithm — not the employer — was making the dispositive screening decision in many cases. The court allowed that theory to be tested.
Workday has continued to deny the allegations and is fighting the class certification. But the practical impact was immediate. HR vendors across the industry began rewriting their customer agreements, adding fairness audit disclosures, and lobbying for state-level safe harbors. The risk that AI screening tools could expose their developers — not just their users — to discrimination liability moved from theoretical to active.
The size of the potential class is the part that should concentrate any CEO’s attention. Workday processes hundreds of millions of applications. Even if the eventual settlement covers only a sliver of the certified class, the exposure runs into the billions.
The Root Cause
Workday sold a screening algorithm as if it were a spreadsheet. The legal architecture treated the software as a passive tool. The actual product made consequential, repeatable decisions that determined whether a human got an interview. When the courts started to look at the function rather than the marketing, the fiction that the employer alone owned the decision collapsed.
The second failure was an absent paper trail. Workday could not produce the kind of disparate-impact testing data that would have been routine in a regulated industry like consumer lending. The lack of documentation became its own evidence that the bias controls had not been built into the system at the depth the marketing implied.
The Pattern to Watch For
If you license an AI tool that makes or recommends consequential decisions about people — hiring, lending, promotion, pricing, insurance — the regulatory frame is shifting from “you, the buyer, are responsible” toward “you and the vendor share liability.” Vendor contracts written in 2022 will not protect you in 2026.
What You Should Steal
Demand from every HR, lending, and personnel-decision vendor the same artifact a regulator would: a documented disparate-impact audit, refreshed quarterly, broken out by protected class, with the methodology disclosed. If the vendor cannot produce it, that absence is your audit finding. Workday’s defense is going to live or die on the documentation it can produce in discovery. Yours will too.