I spend most of my time in this column explaining how Applicant Tracking Systems parse and score resumes so you can work with them instead of against them. This one is different. For the first time, a federal court is being asked a question that used to live only in academic papers and EEOC white papers: when a piece of hiring software rejects you, who is actually responsible, and can you sue them for it? The case is Mobley v. Workday, and as of mid-2026 it's moved from a theoretical question to a real, discovery-stage lawsuit with a nationwide class of potential plaintiffs. Here's what it actually says, where it stands, and, more usefully, what it changes about how you should think about being screened by AI.
What Derek Mobley says happened to him
Derek Mobley is a Black man over 40 who has a disability. Starting in 2017, he applied to more than 100 jobs at companies that use Workday's applicant screening tools and, by his account, was rejected from every single one. What made him suspicious enough to sue wasn't the rejections themselves, job searches are full of rejections, it was the timing. He alleges he sometimes received a rejection notice within minutes or hours of submitting an application, including overnight, patterns that read less like a hiring manager reviewing his resume and more like an algorithm sorting him out before a human ever saw his name. His complaint, filed in the Northern District of California, alleges the screening process discriminates on the basis of race, age, and disability in violation of Title VII, the ADEA, and the ADA. ([Mobley v. Workday case background, FindLaw](https://caselaw.findlaw.com/court/us-dis-crt-n-d-cal/116378658.html))
Where the case stands in the summer of 2026
The case has moved further than most observers expected. On May 16, 2025, the court granted conditional certification allowing the case to proceed as a nationwide collective action under the Age Discrimination in Employment Act, meaning applicants aged 40 or older who were screened through Workday's tools since September 24, 2020, could opt in, and a court-authorized notice process opened to them in January 2026. ([Forbes, "Applied For A Job Through Workday? Court-Authorized Opt-In Is Now Open"](https://www.forbes.com/sites/sheilacallaham/2026/01/13/applied-for-a-job-through-workday-court-authorized-opt-in-is-now-open/)) Then, on June 22, 2026, Judge Rita F. Lin denied most of Workday's motion to dismiss the plaintiffs' latest complaint, including Workday's argument that California's anti-discrimination law shouldn't apply to how it screens applicants outside California. The case is now in discovery, headed toward trial rather than an early exit. ([Akin Gump, "Court Allows Discrimination Claims Against AI Hiring Tool to Proceed"](https://www.akingump.com/en/insights/ai-law-and-regulation-tracker/court-allows-discrimination-claims-against-ai-hiring-tool-to-proceed-or-mobley-v-workday-inc))
The detail that matters more than the verdict
Courts also allowed claims to proceed against Workday as an "agent" of the employers who use its software, a theory that, if it holds, means an AI vendor can be sued directly for discriminatory outcomes rather than hiding behind the companies that deployed it. That's a bigger deal for the industry than any single payout: it means the legal exposure for a biased screening tool doesn't automatically stop at the employer's door.
Can software actually 'discriminate'? The legal test explained
The law doesn't require proving an algorithm was programmed with intent to discriminate, which is good, because almost none are. It relies on a much older concept called disparate impact: a practice is potentially unlawful if it produces a significantly worse outcome for a protected group, regardless of intent. The federal benchmark for "significantly worse" is the four-fifths rule (also called the 80% rule): if a selection process passes members of one group at a rate less than 80% of the rate for the highest-passing group, regulators will generally treat that as evidence of adverse impact worth investigating. ([29 CFR § 1607.4, Cornell Law School](https://www.law.cornell.edu/cfr/text/29/1607.4)) In plain terms: if your screening tool advances 50% of applicants under 40 but only 30% of applicants 40 and older, that 30/50 ratio (60%) falls below the four-fifths threshold and is a legal red flag, even if nobody involved intended it. Mobley's case argues Workday's tools produce exactly this kind of pattern at scale.
The state laws already on the books (and the one that keeps getting delayed)
While Mobley v. Workday works through federal court, a separate and faster-moving track has been happening in state legislatures. A small but growing patchwork of laws already require employers to disclose, audit, or provide notice about AI used in hiring, well before any lawsuit is filed.
| Jurisdiction | What it requires | Status |
|---|---|---|
| New York City (Local Law 144) | Annual independent bias audit of the AI tool, a public summary of results, and 10 business days' notice to candidates before it's used on them | In effect since 2023, but a December 2025 New York State Comptroller audit called enforcement "ineffective" |
| Illinois (Human Rights Act, AI amendment) | Written notice that AI is used in hiring or other employment decisions; AI can't produce a discriminatory effect by protected class or zip code | In effect since January 1, 2026 |
| Colorado (AI Act) | Advance notice before AI is used in a decision, plus 3 years of compliance recordkeeping (originally a much broader risk-assessment law) | Delayed twice and substantially narrowed; now set to take effect January 1, 2027 |
| Federal (EEOC / Title VII, ADA, ADEA) | No AI-specific rule, but existing discrimination law applies to decisions an algorithm makes exactly as it applies to a human's | The EEOC's 2023 AI-specific guidance was quietly removed from its site in January 2025; enforcement continues case by case |
Two things are worth sitting with in that table. First, having a law on the books doesn't mean it's enforced well, New York's own state auditor said so about its own flagship AI hiring law. Second, the trend line only points one direction: toward more disclosure, not less, even as the federal government's own guidance retreated. ([Natural Law Review, "The Federal Government Quietly Removed Its AI Hiring Guidance. Four States Are Writing Their Own."](https://natlawreview.com/article/federal-government-quietly-removed-its-ai-hiring-guidance-four-states-are-writing))
What this actually means for your job search
None of this changes the mechanics I've written about elsewhere on this site, how an ATS actually parses and ranks your resume still works the same way whether or not a lawsuit is pending. What it changes is what you're entitled to ask for, and what's worth documenting.
- ▸If you're rejected within minutes of applying, especially overnight, that's not proof of anything by itself, but it's the exact pattern the Mobley case is built on. Note the date, time, and job title if it happens repeatedly.
- ▸In New York City, you can ask an employer for the AEDT bias audit summary they're required to post, and you're owed 10 business days' notice before an automated tool is used to evaluate you.
- ▸In Illinois, as of 2026, an employer using AI in a hiring decision about you is required to tell you so in writing.
- ▸You're allowed to simply ask a recruiter, "is this stage of your process automated, and is a human reviewing the result before I'm rejected?" A reasonable employer can answer that in one sentence.
- ▸If you're 40 or older and believe you were screened by Workday's tools for a role at any point since September 2020, the collective action's opt-in process is public information you can look up, independent of anything on this site.
It's also worth resisting the instinct to treat every AI hiring tool as an adversary to be tricked. The tools most likely to end up in court are the ones making high-stakes decisions with narrow, poorly audited criteria, not resume-tailoring or scoring assistants you control yourself. Resume Leap's ATS match score works the other way around: it shows you, before you apply, how a parsing and keyword-matching system is likely to read your resume against a specific job description, the same categories a discrimination audit would check, so you can fix a mismatch instead of getting silently ranked out by one.
Key takeaway
AI hiring discrimination went from a theoretical concern to a live, discovery-stage federal case in 2026, and a small but growing set of state laws already give you rights around it, most concretely in New York City and Illinois. You can't opt out of algorithmic screening, but you can ask what's being used, request the notice you're owed where the law provides it, and keep a record of any pattern, like instant or overnight rejections, that looks less like human review and more like a filter.