Duane Morris Takeaway: This week’s episode features Duane Morris partner Jerry Maatman, special counsel Adam Brown, and associate Elizabeth Underwood with their analysis of a significant development in a landmark class action centered on algorithmic discrimination in the employment context.
Check out today’s episode and subscribe to our show from your preferred podcast platform: Spotify, Amazon Music, Apple Podcasts, Podcast Index, Tune In, Listen Notes, iHeartRadio, Deezer, and YouTube.
Episode Transcript
Jerry Maatman: Hello, everyone, and thank you for being here again today for the next episode of the Class Action Weekly Wire. I’m Jerry Maatman, a partner at Duane Morris, and joining me today are my colleagues, Adam Brown and Elizabeth Underwood. Thanks so much for being on our podcast today.
Adam Brown: Great to be here, Jerry.
Elizabeth Underwood: Thanks for having me, Jerry.
Jerry: Today, we’re discussing one of the most closely watched, AI-related employment lawsuits in the United States, a case called Mobley v. Workday. The plaintiffs recently filed a sweeping motion for class certification, asking a federal court in the Northern District of California to allow their claims to proceed on a class-wide basis involving potentially thousands, if not millions, of job applicants who were screened by Workday’s AI-powered recruiting tools. The certification gives us an early preview of the plaintiffs’ case theory and how they’re trying to challenge AI-driven hiring systems on a class-wide basis. Adam, can you give us an overview of the plaintiffs’ motion?
Adam: Sure, thanks, Jerry. The core of this case is an allegation by the plaintiffs that Workday’s recruiting and applicant screening tools disproportionately disadvantage applicants based on protected characteristics, which include race, sex, age, and disability. What they contend is that Workday’s AI products, such as HiredScore Spotlight, Fetch, and Candidate Skills Match, rely on historical hiring data and apply common screening logic across employers, which they say results in systemic disparities. What is important to note here is that the plaintiffs are pursuing disparate impact theories under Title VII, the ADA, and California’s Fair Employment and Housing Act. And what that means is that they don’t necessarily need to prove intentional discrimination, and instead, they’re attempting to show that the tools allegedly produce statistically significant adverse outcomes for protected groups.
Jerry: Well, that’s obviously a very critical decision. President Trump had issued an executive order last year outlawing the use of the disparate impact theory, at least insofar as EEOC actions, and here’s an example where the private plaintiffs’ bar is using that theory to forge ahead on their motion. This isn’t a case, then, challenging individual recruiter-made decisions. Instead, it’s that the technology itself may be generating discriminatory outcomes on a huge scale. That’s what makes this litigation, obviously, so significant for employers and technology providers. Why is this class certification motion noteworthy in and of itself?
Elizabeth: So, the certification motion may ultimately be the most important phase of the case. The plaintiffs are arguing that this dispute presents common questions capable of being resolved in one stroke, which is the language courts often use when evaluating Rule 23 commonality requirements. Their main argument is that Workday implemented the same AI-driven screening architecture across its customer base, so whether those tools produce desperate outcomes should be answered once for everyone subjected to the technology. The proposed class includes applicants who were screened, ranked, scored, shortlisted, or rejected through various Workday AI tools. The plaintiffs also proposed separate subclasses based on race, gender, age, and disability.
Jerry: It’s a very interesting spin in terms of their case architecture and their theories. In other words, the plaintiffs are attempting to move the discussion away from individual hiring decisions and focus it on the design and the operation of the AI systems themselves in an attempt to convince the court that the algorithms work uniformly across the board against employees or applicants in favor of employers. Adam, one aspect of the litigation that’s getting a lot of attention is the statistical evidence cited in the motion. What are the plaintiffs relying on in this situation?
Adam: Yeah, the motion is jam-packed with statistical arguments. The plaintiffs rely very heavily on expert analyses that they contend show significant disparities in rejection and grading outcomes. According to the motion, Black applicants were rejected at rates showing disparities exceeding 18 standard deviations from expected outcomes. Female applicants showed disparities exceeding 15 standard deviations, and Black female applicants allegedly experienced even larger disparities. The plaintiffs also point to alleged patterns in the timing of the rejections. They argue that many applicants receive rejections very quickly after applying, often during evenings, overnight hours, or weekends, and the motion contends that these timing patterns suggest automated decision-making rather than individualized human review.
Jerry: Seems like those allegations in this case theory ties directly into the broader debate surrounding AI in the employment context. The concern is simply not whether a system can make decisions faster. Actually, the concern is whether this automated process is replicating historical patterns in a way that creates headwinds and measurable disparities against protected category groups. Elizabeth, let’s talk about the other side of the “v.” What is Workday’s expected primary responses to this motion likely to be?
Elizabeth: Well, Workday has already previewed much of that argument publicly. The company says the plaintiffs are improperly lumping together different products and different hiring situations. Workday’s position is that applicant outcomes depend on numerous individualized factors, including qualifications, job requirements, employers, industries, locations, and the decisions made by individual customers. Class certification becomes much harder if the court accepts that framework. If every hiring decision requires examining a unique combination of employer preferences and applicant qualifications, common issues may no longer predominate.
Jerry: As we see it happen so often in class-wide litigation, there are competing narratives then underscoring both the motion for class certification and the opposition. Plaintiffs will say, “There’s one AI engine, there’s one common process, and there’s one common question.” I expect Workday is going to argue that there are thousands of employers, millions of applications, and countless individual circumstances, the gist of which is something that doesn’t create a class-wide common question. Whichever narrative the court finds persuasive could determine whether the class proceeds on a class-wide basis or as a collection of smaller individual claims. So, stepping back from that, Adam, what should employers be paying attention to in the coming months with respect to this briefing and the ultimate decision by the court?
Adam: This case is really important to keep an eye on, because it ultimately could be a roadmap for future AI litigation. The theory the plaintiffs are putting forward here is designed to reach beyond a single software platform. They’re effectively arguing that an AI developer can be challenged based on the way its algorithm operates across many employers using a common system ad if courts accept that theory, plaintiff’s lawyers may focus increasingly on centralized AI products rather than individual employers. That could significantly expand the scope of potential AI-related employment litigation.
Elizabeth: And I think another takeaway is the growing emphasis on bias audits and validation studies. One theme throughout the motion is whether Workday adequately tested its systems for disparate impact and whether those efforts were sufficient. Much of the future litigation landscape may focus not just on AI outputs, but also on the governance processes surrounding those systems.
Jerry: These are great points, and the legal conversation seems to be increasingly shifting from “Do you use AI?” to “How do you monitor it, how do you validate it, how do you document those efforts to mitigate one’s risks?” Well, as we wrap up, this remains one of the preeminent and consequential AI employment-related cases in the country, and the court’s upcoming ruling on class certification could have implications well beyond just this litigation. As Adam had indicated, it may well define how future plaintiffs challenge AI -driven employment systems and how courts evaluate class certification in this space.
So, Adam and Elizabeth, thanks so much for joining us today and providing your thought leadership and your insights. And thank you to our listeners for tuning in for another episode of the Class Action Weekly Wire. We’ll certainly keep you apprised of all developments in this litigation.
Elizabeth: Thanks for having me, Jerry, and thank you listeners.
Adam: Thanks, appreciate everyone listening.