Why California Law Just Went Global for HR Tech

Key quote:

Holding business-entity agents like Workday liable for their own discriminatory conduct within the scope of the FEHA is consistent with public policy and the “remedial purposes” of the FEHA.

Why it matters:

The latest order in Mobley v. Workday has some substantial precedents (assuming that they’re not overturned in some future appeals). Judge Rita Lin’s order allowing in part and denying in part Workday’s prior motion to dismiss resolves a couple big questions.

First up, the whole thing where Workday’s “eating their own dogfood” (or “drinking their own champagne”, depending on if you’re from Redmond or San Francisco) would have resulted in liability – that’s gone. For context, if you haven’t been tracking this closely, after the plaintiffs found out that Workday’s secret sauce was covered under ACP, they’d claimed that because Workday used Workday to screen future employees (of Workday), that Workday would be liable as an employer. Because this was added in a later amendment, it was a bit wobbly to begin with. The court called it a new claim that wasn’t in the original complaint, so it wasn’t authorized.

Second, the court denied Workday’s argument that their liability would be based on their customer’s liability. In other words, if an out-of-state employer hiring non-California employees wouldn’t be subject to California law, Workday said neither should it. The judge’s ruling was based on the California Supreme Court’s Raines decision, meaning that Workday can be held liable under California’s Fair Employment and Housing Act (FEHA). This matters because it decouples vendor liability from employer liability. An out-of-state company might never face FEHA claims itself, but the California-based vendor processing applications could still be on the line for how its algorithms operate.

Speaking of California, Bay Area SaaS providers of ADMT tools really aren’t going to be happy about this case. The court found that designing, developing, and operating AI screening tools from California is enough to mean that FEHA applies, even when the rejected applicants live and work out of state. This is different than the second big point, which was around liability. In practice, the location of the server room and the developers in Pleasanton, CA, is enough to pull California law into a case involving applicants in Ohio or New York, something I’d previously noted.

Beyond jurisdiction, the ruling validates a specific theory of disability discrimination that many compliance teams ignored until now. Judge Lin allowed Jill Hughes’s claim to proceed because her complaint detailed how algorithms flag “proxy indicators of illness,” such as medical leave patterns or recovery timelines, rather than just overt health status. HR executives need to know that employment gaps are no longer just a resume red flag; they’re now potential legal evidence of algorithmic bias if a vendor’s system flags them from a California hub.

The procedural losses here offer a roadmap for defense counsel, too. Plaintiffs lost FaithLinh Rowe’s race discrimination claim because they tried to slip in an extra theory for Asian American applicants without asking for leave first, and the court saw right through the attempt to mirror existing allegations. While substantive defenses are getting harder, this proves that normal procedures still help against overreach, since any attempt to expand a complaint beyond what a judge explicitly signed off on should normally result in those excess claims being dropped.

For vendors outside the Bay Area, the danger isn’t just the cost of defense; it’s the unpredictability of jurisdiction regarding choice-of-law conflicts. The court explicitly refused to address these conflicts at this stage, saving that fight for summary judgment or class certification, which means companies can get dragged into immediate discovery and litigation costs in a venue where local statutes apply to global actions. With the clock ticking toward June 29, 2026, when a new version of the Third Amended Complaint must be filed, every vendor using machine learning for hiring needs to have a quick peek at their data pipelines for physical origins, not just legal paper tigers.