Federal Judge Rules AI Hiring Vendor Must Face Bias Claims

Federal Court Rejects 'Just the Tools' Defense

A federal judge in San Francisco ruled on June 22 that Workday, a major AI hiring software provider, must answer discrimination claims brought by Derek Mobley, a Black applicant over 40 who lives with anxiety and depression. Mobley applied to over a hundred jobs and received rejection notices—often in the middle of the night—with such speed and consistency that he suspected no human reviewed his applications. Rather than sue the employers who rejected him, Mobley sued Workday itself in 2023.

Judge Rita Lin rejected Workday's argument that it merely provides tools while employers make the actual hiring decisions. The company claimed its software evaluates job qualifications rather than protected traits like race, age, or disability. Lin allowed discrimination claims to proceed, noting that Workday builds, trains, and operates these tools from its California headquarters, making it subject to the state's anti-discrimination laws regardless of where rejected candidates live.

The lawsuit alleges Workday's software screens candidates using proxies—employment gaps and patterns resembling recurring medical leave—that disproportionately filter out protected groups. No company manual explicitly says "reject cancer survivors" or "reject women of childbearing age," yet proxy measures have long achieved these outcomes. AI has simply automated these patterns at scale.

Why it matters

This ruling establishes that vendors who build and operate AI hiring systems can be held accountable for discriminatory outcomes, not just the companies deploying them. With more than 80% of U.S. employers and nearly every Fortune 500 company now using AI-powered applicant tracking systems, the decision forces a reckoning with the biased historical data these tools learn from. Organizations can no longer treat hiring algorithms as black boxes—they must audit the data feeding these systems and demonstrate that outcomes hold up across protected groups, or face legal liability for scaling discrimination they previously executed slowly enough to escape scrutiny.

AI Didn't Invent Bias—It Made It Trackable

The case highlights a uncomfortable truth: AI hiring tools don't create new forms of discrimination. They're trained on decades of real hiring data that reflects who advanced and who stalled in organizations. The algorithms learn to reproduce that reality without the inconsistency of human judgment.

What's changed is visibility. A decade ago, someone with Mobley's background might have been rejected by a hundred employers through a hundred separate decisions made by different people in different offices. Spread across time and departments, no pattern emerged. One system applying consistent logic to everyone instantly makes discrimination trackable.

Women run roughly 11% of Fortune 500 companies—a figure that actually declined this year. Around ten Fortune 500 CEOs are Black in a country where Black Americans represent roughly one in seven people. These outcomes reflect decades of biased decisions that organizations called "gut instinct" or "culture fit." AI simply automated what was already happening.

The Operational Fix Nobody Wants

Reducing bias requires unglamorous operational discipline: clean data at the source and implement reviews that catch problems where they appear, not years later in depositions. Organizations already audit financial controls and monitor cybersecurity with rigor and budget. The same discipline can apply to hiring.

If an applicant tracking system automatically rejects thousands of candidates annually, companies should be able to demonstrate with data that outcomes hold up across protected groups. The alternative—building what amounts to a caste system that filters out diverse perspectives—doesn't just harm excluded candidates. It bleeds talent, bakes poor instincts into future decisions, and creates products nobody asked for.

The details of the Workday ruling were first reported by Aparna Rae in Forbes.