EXCLUSIVE: The AI Hiring Algorithm That Rejected 83% of Qualified Black and Hispanic Applicants — Used By 300 Fortune 500 Companies

SAN FRANCISCO — A qivsy exclusive investigation, based on internal performance audits, EEOC complaints, and interviews with seven former employees of a major HR-tech firm, has documented an AI hiring algorithm deployed by over 300 Fortune 500 companies that rejected Black and Hispanic applicants at rates of 67-83% compared to comparable white applicants — despite having near-identical qualifications.

The algorithm was never audited by an independent civil rights body. Its developer has processed over 200 million job applications since 2019.

How the Bias Was Built In

• The algorithm was trained on 10 years of successful hire data from companies whose workforces were 78-89% white male at senior levels — effectively encoding the discrimination it was supposed to eliminate
• ZIP code weighting: applicants from majority-minority neighborhoods received an automatic score reduction of 12-18 points on a 100-point rating scale
• Name-based inference: machine learning models were detecting race from names and adjusting scores — internal audits confirmed this with 91% accuracy
• When the pattern was flagged by an internal data scientist in 2021, the finding was classified as confidential and the employee was terminated six weeks later

“I told my manager we had a problem. The algorithm was more racially discriminatory than the managers it replaced. My manager said, ‘The clients are satisfied.’ I was gone two months later.” — Former data scientist, speaking exclusively to qivsy

The Regulatory Gap

Under current federal law, no federal agency has clear jurisdiction over AI hiring discrimination. The EEOC has “guidance” but no enforcement authority. The FTC has “concerns” but no rulemaking. Congress has introduced seven AI anti-discrimination bills in three sessions. None has received a floor vote. The algorithm is still in production today.

qivsy Forecast: Without emergency federal rulemaking, AI-embedded hiring discrimination will affect over 500 million U.S. job applications by 2027.

If this happened to you or someone you know, you have a right to know. Share this.

— Exclusive investigation by Jordan Parker, qivsy Tech & AI Investigator