Erin Kistler applied for thousands of jobs over four years at companies like PayPal, Microsoft, and Netflix. Nearly 20 years of product management experience. Zero interviews. She is now suing Eightfold AI, the Silicon Valley vendor whose hiring software is used by hundreds of employers, in a class-action lawsuit filed in January in California court. The legal theory is novel: automated screening functions as an undisclosed consumer report, ranking applicants without letting them see or challenge the results. The case is one of several converging lawsuits. Workers are suing Meta over an internal AI system that allegedly targeted employees for layoffs because they took parental or medical leave. A separate suit against IBM alleges AI tools discriminated against older workers. All three companies deny the allegations. But the lawsuits share a structural feature: decisions made by opaque algorithms, with no disclosure requirement and no appeals process. The scale is staggering. Last year, 90% of employers used some form of automation in hiring, per the World Economic Forum. Eightfold AI maintains a database of over a billion workers, constantly refreshed from résumés, LinkedIn profiles, and social media. It scores each applicant 0 to 5 on predicted job performance. Research from the algorithmic monoculture study — the largest empirical study of its kind, covering 3.4 million applicants across 156 employers — found that 30% of Black applicants applied to at least one position demonstrating adverse impact against them, and that if Asian applicants were selected at the same rate as the most-selected racial group per position, 29,000 additional Asian applications would have been recommended. Of applicants submitting four applications, 10% were systemically rejected — a rate far exceeding the baseline expected under independent decisions (χ² = 18,481, p < 0.001). The monoculture problem is the structural core. Over 60% of the Fortune 100 use HireVue's algorithms alone. When the same vendor's model rejects you at Company A, it remembers that decision at Company B. Ifeoma Ajunwa of Emory University calls it being "algorithmically blackballed." Katie Creel's research on algorithmic monoculture found that applicants need to submit 25 applications to reach 99.9% probability of at least one recommendation — compared to just 10 under independent decisions. The system doesn't just replicate human bias; it correlates it across the entire market. The bias evidence is damning. Xuechunzi Bai at the University of Chicago ran experiments where AI models made hiring decisions about fictional applicants assigned to made-up demographic groups. After a few rounds, models began stereotyping: if one "Tufa" was a good doctor, all Tufa candidates were favored for doctor roles. The study found greater bias in AI than in human decisions, and newer, more advanced models produced more bias, not less. Amazon previously discontinued an AI hiring tool that downranked women's résumés because historical top performers were men. Ajunwa's research found AI scored applicants with southern accents poorly during voice interviews. The regulatory landscape is patchy. No federal law requires employers to disclose AI use in hiring. New York City's 2023 law mandates annual bias audits and candidate notification, but only for software that "substantially assists" or replaces decisions — a loophole for human-in-the-loop setups. Illinois and Colorado prohibit AI tools that cause unlawful discrimination. Incredible Health, used by 1,500 healthcare employers, audits 10% of its AI interviews for bias. Its CEO, Iman Abuzeid, says AI should collect information for humans, not make decisions — a position that sounds reasonable until you notice it's the exception, not the norm. The legal battles will determine whether algorithmic hiring dossiers get the same transparency regime as credit reports — where consumers can see and dispute their scores — or remain permanently opaque. The plaintiffs' attorneys at Outten & Golden argue for the credit-report model. The vendors argue merit. What neither side disputes: a billion people are being scored, and almost none of them know it.
