Mobley v. Workday Is a Warning About Opaque AI, Not AI Itself

Workday has been ordered to turn over a list of every company that uses its AI-powered hiring tools. That detail, buried in the latest developments in Mobley v. Workday, tells you more about where this case is headed than the arguments about agency law or extraterritorial reach. The court is doing more than deciding whether one vendor can be sued. Our read is that a court does not order a full customer list turned over unless it expects the exposure to reach further than one company. The case has not been decided on the facts, and treating it as a verdict on any single product would be premature while it remains in litigation. Its real significance lies elsewhere: this is not a warning about using AI in hiring. It is a warning about using AI that makes decisions no one inside the hiring company can explain or defend.

From Legal Theory to a Live Class Action

The lawsuit, filed in February 2023, alleges that Workday’s AI-driven HR tools, including HiredScore AI and Candidate Skills Match, discriminate on the basis of race, age and disability in violation of federal and California law. Candidate Skills Match scored and ranked individual applicants, and the plaintiffs allege the system produced discriminatory outcomes at scale across hundreds of employer-customers. Those are allegations. Workday has contested the legal theories underpinning them, and the underlying claims remain unproven.

What has changed is the posture. The court has conditionally certified an Age Discrimination in Employment Act class covering applicants aged 40 and older whose applications were processed by Workday’s AI recommendation system from September 2020 to the present. It has also allowed a claim under California’s Fair Employment and Housing Act to proceed, extending that claim to cover Workday’s tools used by non-California employers evaluating non-California applicants. Attorneys at Womble Bond Dickinson, writing in HR Executive, describe this as the point where courts start reaching the merits of AI discrimination claims rather than dismissing them on procedural grounds. They expect it to open the door to similar suits. That signal is worth taking seriously regardless of how this particular case resolves.

The Court Targeted the Mechanism, Not the Logo

The most consequential holding has nothing to do with the Workday brand. It turns on how the court reasoned its way to letting the FEHA claim stand. Workday argued its liability as an agent should rise or fall with whether its employer-customers are themselves liable. The court rejected that, finding it “at odds with the FEHA’s concept of agency liability.” It held that Workday can be directly liable for its “own engagement in FEHA-regulated activities on the employer’s behalf,” independent of any single customer’s exposure. Workday also argued FEHA should not reach conduct based solely on its California operations. The court disagreed, reasoning that because the tools are designed, developed and operated from California, the alleged discriminatory conduct occurred there regardless of where the applicant sat.

In our view, strip away the choice-of-law reasoning and what is left is a court willing to treat an automated scoring and ranking system as the site of the alleged harm, not the employer’s decision to use it. That is the part every TA leader should sit with. In our view, the exposure sits with mechanisms that score and filter candidates on their own, not with AI that hands a human evidence to weigh. A tool that generates a rationale a recruiter reviews and can defend is a fundamentally different object from a tool that filters candidates out before a human ever sees them.

Vendor Liability Does Not Make You a Bystander

It is tempting to read all of this as Workday’s problem alone. The attorneys’ analysis is explicit that it is not: both the owner of an AI-powered hiring tool and its business customers can be held liable under FEHA. If the allegations against Workday hold up, employers who used its AI recommendation system since 2020 carry exposure alongside the vendor, not behind it. That single fact should end any lingering idea that a signed vendor contract moves the legal risk of an opaque hiring decision somewhere else. It does not. Under this court’s FEHA reasoning, the buyer of the tool is treated as a participant in how it screens people, not a passive customer.

The Steps That Do Not Wait for a Verdict

The court declined to reach Workday’s constitutional arguments, leaving them for a fuller factual record. This litigation still has a long runway left, and waiting for a final ruling before acting is not a defensible position. The mitigations recommended in the attorneys’ analysis, from written AI use policies and vendor due diligence questionnaires to negotiated audit rights, bias testing, disclosures and ongoing human oversight of AI outputs, do not depend on how this case resolves.

The test worth applying is narrower than a full compliance review. Walk your hiring stack and ask, tool by tool: does this system produce a documented, human-reviewable rationale for every candidate it screens, or does it just produce a number? If nobody on your team can open a rejected application and explain in plain language why the system ranked it where it did, you are not looking at an AI efficiency gain. You are looking at exactly the kind of opaque decision-making this case has put in front of a court.

Original reporting: HR Executive.