AI News Deep Dive, July 8: When Your Hiring Algorithm Becomes a Legal Liability

A single federal case has turned AI hiring discrimination liability from a compliance footnote into a boardroom problem. On June 22, 2026, Judge Rita F. Lin in the Northern District of California let most of the claims in Mobley v. Workday move forward, including state-law bias claims that reach the software vendor itself. So the question is no longer whether an employer can be sued for a biased hiring algorithm. It is now whether the company that built the algorithm can be dragged in too. For founders and HR leaders, that shift changes how you buy, audit, and defend every AI tool in your recruiting stack.

What Happened: The Vendor Is Now on the Hook

Derek Mobley, the lead plaintiff, says he applied to more than 100 jobs through employers using Workday’s screening tools and was rejected every time. He is Black, over 40, and lives with anxiety and depression. His argument: the AI ranked him out on those protected traits. (Source: HR Dive)

The case started in 2023. Then it escalated. In May 2025, Judge Lin granted conditional certification of a nationwide collective action under the Age Discrimination in Employment Act, covering applicants 40 and older. (Source: SHRM)

The scale is hard to ignore. In court filings, Workday disclosed that its tools rejected roughly 1.1 billion applications during the relevant period. The opt-in window for the age collective closed on March 7, 2026, so the membership is now set, and the case has entered its merits phase. (Source: Warden AI)

Why does one case matter this much? Because it is the first to test these questions at scale in a US court. The rulings so far are procedural, not a final verdict. Workday denies wrongdoing and will fight on. Still, the direction of travel is clear. Discovery will now pry open how these tools actually score people, and every HR-tech buyer should read the outcome closely.

Why AI Hiring Discrimination Liability Reaches Past the Employer

Here is the part that should get your attention. Historically, anti-discrimination law targeted the employer who made the hiring decision. The vendor selling the screening software sat outside the blast radius. However, Judge Lin’s rulings chip away at that comfort.

In March 2026, she rejected Workday’s motion to dismiss. She also held that the ADEA protects job applicants, not just people already on payroll. (Source: HR Executive) Then in June, she let claims proceed under California’s Fair Employment and Housing Act and the Americans with Disabilities Act. Her reasoning came down to geography. Workday is headquartered in California. Its AI tools were designed and maintained there. So plaintiffs showed a sufficient nexus to apply California law, even to candidates who applied from other states. (Source: HR Dive)

So a candidate in Texas, applying to a job in Ohio, may be able to sue a California vendor under California law. As a result, AI hiring discrimination liability now stretches well past the employer. It puts every algorithmic hiring platform on notice. (Source: Startup Fortune)

Picture what this means for a lean startup. You are a 60-person company. You bought an AI screening tool to save your two recruiters from a resume flood. You never wrote the model. You never saw its training data. But you own the reject decisions it makes on your behalf. And now the vendor who built it may be a co-defendant, not a shield. That is a very different risk profile than the one you signed up for.

Under the Hood: The “Agent” Theory

The legal mechanism matters, because it decides who else gets pulled into cases like this. Plaintiffs argue that when an employer hands screening decisions to a vendor’s AI, that vendor acts as an “agent” of the employer. Under that theory, the vendor takes on the same anti-discrimination duties the employer has.

Why does this land now? Because modern screening tools do more than store resumes. They score, rank, and filter candidates before a human ever looks. When the software effectively makes the reject decision, courts are increasingly willing to treat the toolmaker as a decision-maker, not a neutral utility.

Meanwhile, the disability angle sharpens the risk. Mobley’s ADA claim survived, which means questions about how AI tools handle candidates with mental-health conditions are now in active discovery. For any vendor, discovery is the scary part. It forces disclosure of how the model scores applicants, what data trained it, and whether anyone tested it for disparate impact.

What HR Leaders Do Monday

You do not need to rip out your ATS. But you do need to treat AI hiring discrimination liability as a shared risk between you and your vendors. Start with three moves this week.

Audit before your vendor’s lawyers do

First, ask every hiring-tool vendor a direct question: have you run a disparate-impact audit, and will you share the results? If they dodge, that is a signal. Document what you asked and what they answered. For example, New York City already requires bias audits for automated hiring tools, and the EU AI Act classifies recruitment AI as high-risk, so this paperwork is becoming table stakes anyway.

Keep a human in the loop, on paper

Second, make sure a person reviews borderline rejections, and record that they did. The weakest legal position is a fully automated funnel with no human checkpoint. A documented review step is cheap insurance. It also improves your hiring, because algorithms miss context that a recruiter catches in ten seconds.

Rewrite your vendor contracts

Finally, push for indemnification language that covers algorithmic bias claims. If your vendor built the model, they should carry part of the risk. Loop in employment counsel before you renew, because contract terms signed today decide who pays tomorrow. Teams that run AI in HR recruitment without these guardrails are the ones most exposed.

The bigger picture is simple. AI can screen faster, but speed without auditability is now a liability, not an edge. Therefore, if you are building your process from scratch, design the hiring policy and the audit trail together. Asanify’s applicant tracking system keeps a clear record of every stage, which is exactly the documentation these cases turn on.

FAQ: AI Hiring Discrimination Liability

Can a software vendor be sued for AI hiring bias?

Yes, increasingly so. In Mobley v. Workday, a California federal judge allowed bias claims to proceed against the vendor under an “agent” theory, where the AI tool acts on the employer’s behalf. This means the company that builds the screening algorithm can share legal exposure with the employer using it.

Does this ruling affect employers outside California?

It can. Judge Lin held that because the vendor’s AI tools were designed and maintained in California, state civil rights law may apply even to applicants who applied from other states. So employers using California-built hiring tools should assume broader exposure and audit their screening process accordingly.

How do I reduce AI hiring discrimination liability at my company?

Run disparate-impact audits on any automated screening tool, keep a documented human review step for rejections, and negotiate indemnification for algorithmic bias in vendor contracts. In addition, consult employment counsel before renewing AI hiring software. These steps create the audit trail that regulators and courts now expect.

Not to be considered as tax, legal, financial or HR advice. Regulations change over time so please consult a lawyer, accountant  or Labour Law  expert for specific guidance.