A recent post on X by prediction market Polymarket has brought renewed attention to a Stanford study indicating that AI hiring tools disproportionately screened out Black and Asian applicants. The study, originally published nearly a month ago, resurfaced after the post drew widespread engagement.
The study was based on a large dataset of 4 million job applications to 156 different employers in 11 business sectors. This allowed the researchers to quantify just how skewed the automated platforms, which are used in recruitment by nearly 75 percent of all global companies, can be and how they risk becoming worse over time.
“We find substantial evidence of racial disparities in AI-based candidate screening,” the authors of the study wrote in an article on Stanford’s website. “We discovered that 26 percent of Black applicants and 15 percent of Asian applicants applied to positions where the AI system discriminated against their racial group.”
READ: Workday to face lawsuit over hiring bias in AI-driven HR software (June 23, 2026)
The study mentions that this hiring bias would have practical consequences for candidates AI tools are prone to reject that goes beyond denials of jobs they might be qualified for. Since these platforms are used by different employers, Black and Asian applicants could face unfair rejections from entirely different companies as well.
“As a single hiring vendor comes to dominate screening for an industry, it may be more likely that candidates are shut out,” the researchers said. “When applying to two positions at two different employers, applicants might reasonably expect that they are receiving two separate evaluations and therefore two chances. But if both positions share the same model, their numerical score will be identical.”
The study largely analyzed decisions made by the popular hiring tool Pymetrics, which screens applicants based on responses to online games they’re asked to play. While the ultimate decision on whether to recommend a candidate lies with the employer, researchers found evidence that the platform proposed Black and Asian qualifiers at disproportionately lower rates than other cohorts.
“To put this in perspective: If the AI had recommended Black and Asian candidates at the same rate as it recommended the most-favored group (typically white applicants), 40,000 more of their applications would have advanced to the next stage of hiring,” the researchers said.
READ: Nvidia tackles tech layoffs with high-paying AI hiring (June 2, 2026)
They added the degree of bias met the Equal Employment Opportunity Commission’s “adverse impact” definition: “when one group is recommended at less than 80 percent of the rate of the most-recommended group.” As a result, the study concluded that the AI screening tool found ways of distinguishing the race of applicants even in the absence of that detail being specified.
This comes amid wider concerns about the issue. A U.S. District Judge recently ruled that human resource management platform Workday must face claims that its popular AI-powered human resources software weeded out job applicants at other companies in ways that violated California law and a federal ban on discrimination against workers with disabilities.