Employers Are Now Legally Liable for What Their AI Hiring Vendor Does. Here Is What That Means.

In February 2026, a federal court in California authorised class-action notice in Mobley v. Workday: a lawsuit that has been closely watched by the HR technology industry since it was filed in 2023. The court's ruling on the liability question is the most significant development in AI hiring law to date.

The court treated Workday as an "agent" of the employer in the hiring process. This means the employer, not the software vendor, bears primary legal responsibility for discriminatory hiring decisions made by the AI tool, even when the employer had no direct involvement in how the algorithm was configured or trained.

What the Ruling Actually Says

The plaintiff in Mobley v. Workday, Derek Mobley, alleged that Workday's AI hiring platform discriminated against him on the basis of race, age, and disability. Workday's defence argued that it was a software vendor, not an employer, and therefore could not be held liable under employment discrimination law.

The court rejected this argument. By functioning as an active participant in the hiring decision-making process, screening candidates, ranking applicants, and making recommendations that directly influenced which candidates were considered, Workday was acting as an agent of the employers using the platform. Agents of employers are subject to the same anti-discrimination obligations as employers themselves.

The age discrimination claims under the Age Discrimination in Employment Act (ADEA) were allowed to proceed. This is particularly significant because age discrimination in AI screening is one of the most common concerns raised by employment law practitioners: training data that over-represents recently graduated candidates can systematically disadvantage older applicants in ways that are difficult to detect and trace.

The Eightfold AI Companion Case

The Mobley ruling was preceded by a separate proposed class action against Eightfold AI, filed in January 2026. The Eightfold complaint alleges that the platform compiles personal data, LinkedIn profiles, job histories, public professional records, to rate candidates for roles without obtaining the candidates' consent, in violation of the Fair Credit Reporting Act (FCRA).

Together, the Mobley and Eightfold cases establish two distinct legal risk vectors for AI hiring tools:

The Mobley vector: discrimination liability for algorithmic decisions that systematically disadvantage protected classes. The employer cannot separate their legal exposure from the vendor's algorithm.

The Eightfold vector: privacy liability for data compilation practices that constitute consumer reporting without FCRA compliance. The platform cannot freely aggregate public data about candidates without legal obligations to those candidates.

What This Means for Recruitment Agencies

The immediate relevance to recruitment agencies is in how they advise clients on technology adoption and how they use AI screening tools in their own processes.

Advising clients on ATS selection. Many recruitment agencies assist their clients in selecting and implementing ATS and screening tools. The Mobley ruling means that part of this advisory service should now include a legal risk assessment of the tool: has the vendor conducted bias audits, how are protected class variables handled in the algorithm, and what is the vendor's position on employer liability for discriminatory outcomes? The agency that fails to raise these questions and a client subsequently faces discrimination litigation may face its own liability questions.

Using AI screening in the agency's own process. Agencies that use AI tools to screen candidates before presenting shortlists to clients are in the same legal position as Workday. If the agency's AI screening tool systematically disadvantages candidates based on protected characteristics, the agency, not the tool vendor, bears primary liability for the discriminatory placement practice. This is the same "agent" logic the Mobley court applied.

Documentation of human oversight. The practical defence against AI discrimination liability is documented evidence that humans exercised meaningful oversight of algorithmic recommendations. Agencies that can show the AI generated a recommendation, a human recruiter reviewed the recommendation against defined criteria, and the human made the final selection decision are better positioned than agencies where the AI recommendation effectively is the decision.

The EU AI Act Context

The US litigation development is occurring against the backdrop of the EU AI Act, which classifies AI tools used in recruitment, candidate selection, performance monitoring, promotion, and termination as "high-risk AI systems" subject to mandatory compliance requirements. The EU delayed these requirements from August 2026 to December 2027 through the Digital Omnibus agreement reached on 7 May 2026.

The delay gives Australian agencies that operate with EU clients additional time before formal compliance obligations apply. But the underlying legal logic of the Mobley ruling, that AI tools exercising hiring judgment are agents subject to anti-discrimination law, is not jurisdiction-specific. It is a fundamental framing of how courts understand AI decision-making in hiring contexts.

Australian employment law does not yet have an equivalent ruling. The Fair Work Act and the Racial Discrimination Act create comparable anti-discrimination obligations. As AI screening tools become standard practice in Australian recruitment, it is a matter of when, not whether, similar litigation reaches Australian courts.

The Practical Response

For recruitment agencies using AI screening tools, three practices reduce legal exposure:

Require bias audit documentation from all AI screening vendors before adoption. Reputable vendors will have third-party audit reports available. Vendors that cannot produce bias audit documentation should not be used in a hiring context.

Document the human oversight layer explicitly. Every AI recommendation should be reviewed by a named human recruiter who applies defined criteria before the recommendation influences a placement decision. The documentation of this review is the evidentiary record that distinguishes "AI-assisted" from "AI-determined" selection.

Review candidate data compilation practices against FCRA principles, the same principles that govern employment background checks, for any tool that aggregates candidate data from public sources. Data compilation that would require FCRA compliance if done by a credit bureau does not lose that obligation because it is done by an AI tool.

The Mobley ruling is a landmark in a development that is still early. The legal framework for AI hiring accountability is being built in real time by courts responding to real plaintiffs. The agencies that build compliance practices now are building defensible positions for the litigation environment that is coming.