What is Mobley v. Workday about?

Mobley v. Workday is a federal case in which a job applicant alleged that Workday's AI-powered applicant-screening tools discriminated against candidates based on age, race, and disability. The novel question is not whether an employer can be liable — that's settled — but whether the AI vendor itself can be sued for employment discrimination. The court allowed claims to proceed on the theory that a screening vendor can act as an employer's 'agent,' and later allowed the age-discrimination claim to advance as a nationwide collective action.

Can an AI hiring vendor be liable for discrimination, not just the employer?

That is exactly the theory being tested. Traditionally, anti-discrimination statutes like Title VII and the ADEA target 'employers.' Mobley advances the argument that when a vendor's AI performs a core hiring function — screening, ranking, or rejecting candidates on the employer's behalf — the vendor is acting as the employer's agent and can face liability. If this theory holds broadly, AI hiring vendors face direct exposure, not just reputational risk.

What is disparate impact and why does it matter for AI hiring?

Disparate impact is discrimination that results from a facially neutral practice that disproportionately harms a protected group, even without intent to discriminate. AI hiring tools are especially exposed because a model trained on historical data can systematically disadvantage older workers, certain races, or people with disabilities while appearing neutral. A plaintiff generally doesn't need to prove the vendor meant to discriminate — a statistically adverse effect on a protected class can be enough to state a claim.

What should employers and HR-tech vendors do in response to Mobley?

Employers should demand bias-audit results before adopting an AI hiring tool, keep a human in the loop for consequential decisions, and negotiate clear indemnification. Vendors should run and document adverse-impact testing, avoid features that proxy for protected characteristics, provide transparency and audit support, and revisit contract terms that assume all discrimination liability sits with the employer. Both sides should track state AI-hiring laws (e.g. NYC Local Law 144, Colorado's AI Act, Illinois) that add audit and notice duties.

Mobley v. Workday: Can an AI Hiring Vendor Be Liable for Discrimination? (2026)

What Happened

The plaintiff applied to a large number of jobs through employers that used Workday's AI-powered applicant screening and was rejected repeatedly. He alleged the tools discriminated on the basis of age, race, and disability — and, unusually, sued the software vendor rather than (only) the employers.

Two developments made the case matter beyond its own facts:

The court declined to dismiss the discrimination claims against the vendor, letting the theory that the vendor could be liable as the employers' agent move forward.
The court permitted the age-discrimination (ADEA) claim to proceed as a nationwide collective action — dramatically expanding the potential class of affected applicants.

None of this is a final finding that Workday discriminated. But allowing the claims to advance signals that courts are willing to entertain vendor liability for algorithmic hiring outcomes — which is precisely the exposure HR-tech companies had assumed sat only with their customers.

Why the "Agent" Theory Is a Big Deal

Statutes like Title VII and the ADEA regulate "employers" and their agents. The traditional assumption in HR tech was: we sell software; the employer makes the decision; the employer bears the risk. The agent theory challenges that split. If an AI tool is doing the screening, ranking, and de-facto rejecting, a plaintiff can argue the vendor is performing a function that the law attributes to the employer — and therefore shares the liability.

The shift: "We're just a neutral tool provider" is a weaker defense when your tool autonomously decides who advances. The more decision-making authority the AI holds, the more it looks like an agent — and the harder it is to disclaim responsibility for discriminatory outcomes.

Disparate Impact: The Mechanism That Catches AI

AI hiring discrimination usually isn't about intent — it's about disparate impact. A facially neutral model can systematically disadvantage a protected group because it learned patterns from biased historical data. Common failure modes:

Age proxies

Graduation years, 'digital native' signals, or tenure gaps that correlate with age and screen out older workers — the core of the ADEA claim in Mobley.

Race and gender proxies

ZIP codes, names, schools, or hobbies that correlate with protected characteristics even when race/gender aren't inputs.

Disability screens

Gamified assessments, video-analysis, or timing metrics that disadvantage applicants with disabilities and can raise ADA issues.

Feedback loops

Training on 'who we hired before' teaches the model to reproduce past bias, then calls it a prediction of success.

Because disparate-impact claims don't require proof of intent, a statistically adverse effect on a protected class can be enough to open the door — the burden then shifts to justifying the practice as job-related and consistent with business necessity.

What Employers Should Do

Demand the bias audit. Require adverse-impact testing results before adopting any AI screening tool — and re-run them periodically. Some states (e.g. NYC) mandate this already.
Keep a human in the loop. Don't let the AI auto-reject. Meaningful human review of consequential decisions weakens the "the machine did it" problem.
Negotiate indemnification. Push back on vendor contracts that dump all discrimination liability on you. Mobley shows the vendor may be on the hook too.
Document business necessity. Be able to show each screening criterion is job-related — the classic disparate-impact defense.

What AI Hiring Vendors Should Do

Test and document adverse impact across age, race, gender, and disability — before shipping and on an ongoing basis.
Strip proxy features that stand in for protected characteristics (graduation year, ZIP, name-derived signals).
Build transparency and audit support so employer-customers can meet NYC Local Law 144, Colorado's AI Act, and Illinois requirements.
Revisit your contracts. Terms that assume 100% of discrimination risk sits with the employer no longer match the legal landscape.

Frequently Asked Questions

Has a court found that Workday actually discriminated?

No. Allowing claims to proceed past a motion to dismiss, and certifying a collective action for notice purposes, is not a finding of liability. It means the court accepted that the plaintiffs plausibly stated claims worth litigating. The significance is the theory being allowed to advance — vendor liability via agency — not a final judgment.

Does this only affect large vendors like Workday?

No. The agency theory turns on function, not size. Any vendor whose AI screens, ranks, or rejects candidates on an employer's behalf could face the same argument. Smaller HR-tech and AI recruiting startups arguably have more exposure because they have fewer resources for bias auditing and legal defense.

How does this interact with state AI hiring laws?

It stacks. Mobley is about federal anti-discrimination statutes (Title VII, ADEA, ADA). Separately, states impose their own duties: NYC Local Law 144 requires independent bias audits and candidate notice for automated employment decision tools; Colorado's AI Act imposes duties on developers and deployers of high-risk AI including hiring; Illinois regulates AI video interviews. A vendor can face federal discrimination exposure and state audit/notice obligations at once.

We use AI only to source candidates, not to reject them. Are we exposed?

Lower, but not zero. If your AI meaningfully shapes who an employer ever sees — surfacing some candidates and burying others — it can still produce disparate impact and perform a screening-like function. The key risk factor is how much decision-making authority the tool holds over who advances.

The Takeaway: "Just a Tool" Isn't a Shield Anymore

Mobley v. Workday doesn't settle the question of vendor liability — but it moves the ground under everyone building and buying AI hiring software. The more your AI decides, the more it looks like an agent, and the harder it is to disclaim responsibility for who gets screened out. Bias auditing, proxy removal, human oversight, and honest indemnification terms move from "nice to have" to baseline risk management.

Whether you deploy AI screening or sell it, the defensible position is the same: know your tool's adverse impact, document it, and be able to show each criterion is job-related. Everything else is exposure waiting for a plaintiff.