Background
On June 22, a federal judge ruled that a discrimination lawsuit against Workday, one of the world’s largest people management software providers, could move ahead in California court. The suit, Mobley vs. Workday, Inc., filed with the EEOC in California’s Northern District Court, alleges that Workday’s AI-powered hiring assistance tools are illegally denying candidates belonging to protected classes including age, race, and disability status.
Workday offers teams large and small a vast suite of tools, but this particular case focuses on Workday’s job listing, application, and candidate screening tools. Since, according to Workday’s own reporting, over 11,000 different firms worldwide employ Workday’s services (and over 80% of those users utilize Workday’s AI offerings), questions regarding Workday’s software are wide-reaching and affect millions of managers and jobseekers alike.
Workday (or other comparable web-based job portals) is often the only option given to candidates to apply. Workday also screens, rejects, and advances candidates for clients. Therefore, according to the earliest rulings in this case, federal and state anti-discrimination rules apply to Workday as well as Workday’s clients.
Of course, this specific lawsuit does not exist in a bubble. In 2025 California amended its employment regulations to make employers liable if an AI is found to discriminate against an applicant, and New York City requires employers to publicly audit any AI hiring tools to check for potential bias.
How Workday AI Works, and What Went Wrong
Workday, like many other technology providers, offers a wide range of agentic AI tools to help hiring managers with their candidate search. This includes tools which score candidates based on how well the AI believes they would fit a role, and at times automatically rejecting candidates based on what the AI considers to be non-starters. These AI tools apply to typical applications, as well as assessments and personality tests.
These tools are part of a suite meant to help hiring managers work through high volumes of applicants faster, and find the best candidates to advance in the hiring process easier. However, the individual suing Workday alleges that their AI would automatically, sometimes within minutes, reject them from a position they were otherwise qualified for- because of legally protected characteristics.
In this specific case, the individual suing Workday is black, middle-aged, and has a disability. The applicant believes they are qualified for the roles they applied for, and since they were receiving seemingly automated rejections very quickly after applying, the individual suing alleges that Workday has been automatically rejecting them from jobs on account of their age, race, and disability status.
The root causes for biases appearing during AI screening are not yet fully understood, and are thus very difficult to eliminate entirely. These biases could exist due to a lack of diversity on AI development teams, lower priorities, or a reliance on historical hiring patterns; which have certainly been known to be imperfect at times.
Guiding Your AI Tools and How to Avoid Problems
- Learn from their vendors how their AI tools work, and find out if there are any automated safeguards to protect against bias. Knowing potential shortfalls and problem areas early on will make it easier to solve these problems later in implementation
- Stay involved at every step of the hiring process, and potentially exclude AI entirely from steps in the hiring process involving EEOC rules or other employment fairness laws. With a current lack of legal clarity, human intervention can be a team’s best defense against accidental AI bias
- Be transparent with internal stakeholders and external candidates alike as to when and how AI is used in the hiring process