HR platform Workday functions as a layer between companies and talent. But a lawsuit in a US federal court in California alleges that when that layer screens applicants automatically without human oversight, a single flaw in the system can follow a candidate from one employer to the next. 

According to the lawsuit, Derek L. Mobley, a Black man born in 1974, was rejected for over 100 jobs, often within hours of applying, and sometimes at late hours, such as 2 AM. Those jobs, posted on LinkedIn, Indeed, Monster, or CareerBuilder, had one thing in common: the Workday platform. 

Mobley sued Workday in 2023 (Mobley v. Workday, Inc., No. 3:23-cv-00770, N.D. Cal.). His resume shows that he graduated from Morehouse College in Atlanta with a bachelor’s degree in finance, according to the complaint. Morehouse is an all-male historically Black college and university, which, the lawsuit argues, could serve as a proxy for his race. His extensive employment history, it adds, could serve as a proxy for his age. 

Mobley argues that Workday’s AI filtered out people like him, who never got a chance to be screened by a human. 

The lawsuit claims that, as of April 2023, nearly one in four US job openings was processed on the Workday platform. 

Workday says the product is highly customizable and that customers decide how to source, evaluate, and reject job applicants. 

So, who bears responsibility when AI-powered hiring tools allegedly discriminate against groups? 

In July 2024, US District Judge Rita Lin allowed part of the case to proceed on the theory that Workday can act as an agent of the employers that delegated hiring tasks to its tools. 

“Drawing an artificial distinction between software decision-makers and human decision-makers would potentially gut anti-discrimination laws in the modern era,” the judge wrote. 

In May 2025, the court preliminarily certified a collective action under the Age Discrimination in Employment Act (ADEA) of 1967, on behalf of applicants aged 40 and over who applied for jobs through Workday’s platform since September 2020 and were denied employment recommendations. That is an early step that lets the age claim proceed as a group, not a finding that Workday discriminated. 

As of early November 2026, the case remains pending and is in the discovery phase. The court has not ruled on whether Workday is liable. 

In the European Union, the AI Act classifies AI systems used to recruit or select people as high-risk systems (Annex III, point 4). The obligations for these systems, postponed to December 2, 2027, will fall not only on providers (the vendors who build the systems) but also on deployers (the businesses that use them). High-risk systems will have to allow effective human oversight. 

So, what should you be asking your vendors? 

Again, the EU AI Act’s high-risk obligations won’t apply until December 2, 2027, but employers should be proactive and ask the six questions below before buying an AI screening tool. 

1. Who controls the configuration? 

Ask which settings (such as scoring criteria, knock-out questions, and thresholds) are fixed by the vendor and which are set by your team, and whether changes are recorded. 

If the vendor’s position is that the customer decides, you should know what you’ve decided. 

2. Does the system reject candidates automatically? 

Ask whether any step could end an application without a human looking at it, and whether that step can be switched off. 

General Data Protection Regulation (GDPR) Article 22 gives people the right not to be subject to a decision based solely on automated processing that has legal or similarly significant effects on them, with limited exceptions. 

3. How was the system tested for bias? 

EU AI Act Article 10 covers high-risk systems that use model training. It requires that training, validation, and testing data sets meet quality criteria that include examination for possible biases as well as appropriate measures to detect, prevent, and mitigate them. Ask the vendor to provide details about the training and any biases detected. 

4. Can a recruiter see why a candidate was ranked or rejected? 

Article 13 requires the system to be transparent enough for deployers to interpret its output, and Article 14 requires that people can oversee it effectively. Ask the vendor for the instructions for use, check that they are clear and easy to find, and ask what explanation a recruiter actually sees when a candidate is ranked or rejected. 

5. What is logged, and can you access it? 

Article 12 requires high-risk systems to allow for the automatic recording of events, and Article 26(6) requires deployers to keep the logs they control for at least six months, unless other EU or national law provides otherwise. Ask what the system records, and make sure you can export the logs yourself without going through the vendor. 

6. Who informs the candidate about AI use? 

Article 26(11) requires deployers to inform people that a high-risk system is being used in decisions about them. Article 86 gives affected people the right to an explanation of the system’s role in a decision. Ask whether the vendor provides notice wording and the information needed to answer such requests. 

Don’t let a single filter decide who gets seen 

The Workday case doesn’t show that automated screening is unreliable. So far, the claims remain allegations, and Workday denies them. It does highlight a structural risk, though: that a flaw in such a system goes unnoticed and affects hiring at other companies using it. Such a flaw can be biased patterns in the historical data a tool learns from, in which case the bias predates the tool.

Beyond asking vendors the questions above, another way to partially counter this risk is to keep active headhunting in your hiring process. Recruiters who maintain professional networks and approach candidates directly can reach people who never applied, or whose applications a filter might have ranked low. This channel also strengthens human participation in your recruitment process. 

Of course, networks have limits, too. For example, referral-heavy hiring can reproduce the profile of the existing team. A good principle here is to diversify, so no single channel (whether an applicant tracking system, a referral scheme, or an external recruiter) becomes the only gate. Also, review each channel for who it leaves out. 

If your recruiting team can’t cover active headhunting, consider an external partner. Jobshark does not rely solely on automated tools to decide which candidates employers see. A team of recruiters actively headhunts from a network of hundreds of thousands of IT professionals. 

Wrapping up 

Job seekers often do their best to polish their resumes, network, and stay positive through rejection. They deserve a fair process. Employers also have an interest in hiring on talent, not age or ethnicity, and in complying with local legislation. 

The Workday case hasn’t been decided, and its claims remain allegations. But when a single vendor’s software screens applicants for so many employers, one flawed filter may turn rejection from one company into rejection from every employer using it. 

That doesn’t mean the response is to drop AI tools. But you should keep them in check. Ask vendors the six questions above, keep a qualified person involved in rejection decisions, and don’t let any single channel be the only gate. 

Combining automated screening with active sourcing and networking can help ensure that fewer qualified candidates are filtered out unseen. 

Want to stay informed with content that matters?  

Join 30,000+ HR professionals and tech leaders: subscribe to the TechTalents Insights newsletter and get our best articles and interviews, completely free.