AI Hiring Tools Raise New Concerns Over Racial Bias In Job Searches
As the Class of 2026 enters a tight job market, a new Stanford-led study is raising serious questions about how artificial intelligence is shaping who gets considered for work — and who gets screened out before a person ever sees their application.
The study, published by Stanford’s Institute for Human-Centered Artificial Intelligence, examined millions of real job applications processed through an AI hiring tool. Researchers found that the technology showed concerning racial disparities, particularly for Black and Asian applicants.
According to the study, researchers followed roughly 3.4 million people who submitted 4 million job applications to 1,700 job postings across 150 employers and 11 industry sectors. Each application was reviewed by an AI hiring tool built by one third-party vendor.
Researchers found that 26% of Black applicants and 15% of Asian applicants applied to positions where the AI system discriminated against their racial group under the study’s adverse-impact analysis. The researchers used the Equal Employment Opportunity Commission’s “four-fifths rule,” which flags potential discrimination when one group is selected at less than 80% of the rate of the most-selected group.
The numbers were not small. Researchers estimated that if Black and Asian candidates had been recommended at the same rate as the most-favored group, about 40,000 more applications would have advanced to the next stage of hiring.
The study also found that broad averages can hide discrimination. When all recommendations from the vendor were pooled together, the researchers did not find adverse impact. But when they looked position by position, disparities appeared across many jobs. In plain English: the big spreadsheet can look clean while individual hiring lanes are still dirty.
One AI Vendor Can Mean Rejection Across Multiple Employers
The study also warned about what researchers called “algorithmic monocultures,” where many employers rely on the same few vendors or systems to screen applicants.
That matters because one algorithm used across many companies can repeatedly reject the same candidates. Researchers found that applicants who submitted multiple applications through the same AI vendor were more likely to be rejected from every position they applied to than would be expected if each company made independent decisions.
The paper’s abstract reported that 4% of applicants who applied to 10 positions were recommended for rejection from all of them, a rate higher than chance.
The researchers compared those results with a prior large study of hiring outcomes that did not focus on AI, and they did not find the same pattern of across-the-board rejection. That suggests market concentration — many employers depending on the same tool — may worsen the problem.
Study Adds To Growing Scrutiny Of AI In Hiring
The findings come as AI hiring systems are facing more public and legal scrutiny. Reuters reported this week that Workday must face a California lawsuit alleging its AI-powered hiring software discriminated against job applicants, including claims involving disability-related screening concerns. Workday has denied wrongdoing and says its AI evaluates qualifications while following its responsible AI program.
Stanford researchers said AI hiring tools combine three traits that make them especially risky in high-stakes decisions: they are widely used, deeply consequential and often opaque to the public.
Their conclusion was clear: independent research is needed to understand how algorithmic hiring affects individual job seekers and the broader workforce.
For Black and Asian applicants, the concern is not only whether AI can make hiring faster. It is whether the same technology promising efficiency is quietly becoming another gatekeeper.
