OpenAI just handed over $3.2 million to the Justice Department, settling federal allegations that it systematically favored foreign workers on temporary visas over American job seekers. The company and its subsidiary Statsig Inc reached the deal after investigators found deliberate recruitment tactics designed to limit opportunities for US citizens.
The specifics are damning. Certain positions went unadvertised on public job boards entirely. Late-night postings and paper-only applications specifically deterred American candidates from applying. These weren't accidents or oversights, federal officials argue, but deliberate barriers built into the hiring process itself.
What the Settlement Covers
The deal breaks down into concrete payments. OpenAI pays a $1.2 million civil penalty straight to the government. The remaining $2 million goes into a fund to compensate workers who were actually harmed by the discrimination. The company also faces mandatory training on hiring law and ongoing federal oversight of its recruitment practices going forward.
This marks the 13th enforcement action under a revived Justice Department program targeting citizenship discrimination in employment. The probe examined fewer than ten positions at OpenAI, yet still uncovered enough to justify federal intervention and a nine-figure settlement.
OpenAI's statement reveals the tension between what the company wants and what the law requires. A spokesperson said the firm "disagrees with the DOJ's findings" but agreed to settle anyway to move forward with its PERM program, which lets employers sponsor foreign workers for permanent residency when they claim they cannot find qualified Americans. Federal rules explicitly prohibit favoring foreign workers over qualified US applicants or discriminating based on citizenship status during hiring, yet OpenAI apparently did exactly that.
This article is informational only and not financial or legal advice. Settlement terms and employment law compliance are complex matters that may affect different parties differently.
