AI may be quietly deciding who gets a shot at a paycheck before a hiring manager ever opens their resume, and a new Stanford study says Black applicants, as well as Asian job seekers could be paying the biggest price.
Researchers at Stanford HAI analyzed 3.4 million people who submitted 4 million applications to 1,700 job postings across 150 employers and 11 industries. Every application in the study was reviewed by an AI hiring tool from one third-party vendor, giving researchers what they called a rare look inside the “black box” of algorithmic hiring.
The findings were alarming. According to Stanford, 26% of Black applicants and 15% of Asian applicants applied to jobs where the AI screening system showed discrimination against their racial group. Researchers said that if Black and Asian candidates had been recommended at the same rate as the most-favored group, usually white applicants, 40,000 more applications would have moved forward.
The system worked by labeling candidates as “recommend” or “do not recommend,” then sending that decision to employers. That may sound like a simple screening shortcut, but Stanford’s study suggests the same tool being used across multiple companies can create what researchers call an algorithmic monoculture. In plain terms, if one AI system does not like your resume, it may keep rejecting you everywhere that same vendor is used.
That is especially concerning as the Class of 2026 enters a job market already filled with uncertainty. NACE’s Job Outlook 2026 report found employers projected only a 1.6% increase in hiring for new college graduates compared to the Class of 2025, while many recruiters described the market as only “fair.” For job seekers already fighting through layoffs, slow hiring, and hundreds of online applications, an invisible AI rejection system could make the process feel even more impossible.
Stanford also warned that measuring bias the wrong way can hide the problem. Researchers used the EEOC’s “four-fifths rule,” which flags potential adverse impact when one group is recommended at less than 80% of the rate of the most-recommended group. Looking at all recommendations together may make the system appear neutral, but checking job by job revealed disparities. As Stanford put it, “The big-picture average hides the real discrimination happening job by job.”
The concern is no longer just academic. On June 22, a federal judge ruled that Workday must face claims in a California lawsuit alleging its AI-powered HR software screened out applicants in ways that violated state law and disability discrimination protections. Workday denied the claims, saying, “Our technology looks only at job qualifications, not protected traits like race, age, or disability. We rigorously test our products as part of our Responsible AI program to confirm our tools do not harm protected groups.”
The EEOC has also warned that AI and algorithmic tools may “mask and perpetuate bias or create new discriminatory barriers to jobs.” Stanford’s takeaway was just as direct: these tools are “pervasively adopted, highly consequential, and opaque to the public.”
For workers, the issue is not just whether AI can read a resume faster. It is whether the technology is quietly blocking qualified applicants, especially Black applicants, from ever reaching a human.
