// news · policy2026-08-22source: Governance reporting and regulatory analysis

The Justice Department settled its first AI hiring discrimination case

The Civil Rights Division settled with OpenAI OpCo and Statsig over citizenship-status discrimination in PERM recruitment workflows. It is among the first federal civil rights actions aimed directly at AI-assisted hiring, and it did not need a new statute to happen.

The US Department of Justice's Civil Rights Division has settled with OpenAI OpCo and Statsig over citizenship-status discrimination in PERM recruitment workflows — one of the first federal civil rights enforcement actions to target AI-assisted hiring directly.

Why it matters more than its size

Nothing here required an AI statute. Citizenship-status discrimination in recruitment was already unlawful; the Immigration and Nationality Act's anti-discrimination provision has been enforceable for decades. What is new is the Division applying it to a workflow where a model sits between the applicant and the decision.

That is the enforcement route most likely to reach organisations first, and it is widely underestimated. Firms watching Brussels for the rules that will bind them are watching the wrong jurisdiction if they hire in the United States, because the existing employment, credit and housing statutes already cover automated decisions and already have enforcers.

The specific exposure

PERM workflows are a narrow, documentation-heavy process, which is precisely why an automated step is attractive and precisely why it leaves a trail. A recruitment system that filters on a protected characteristic does not become lawful because a vendor supplied it, and "the model did it" has never been a defence in any other regulated decision.

What to check

Any automated step in hiring, promotion or termination. Whether the filtering criteria are documented and defensible. Whether the vendor contract makes clear who answers for the output. And whether anyone in the organisation could reconstruct, a year later, why a particular candidate was screened out.

AI Governance Weekly — AI Governance Weekly — August 20, 2026 → · EveAI Core — AI Regulation in 2026: What Just Changed → · Holland & Knight — US Companies Face EU AI Act's August 2026 Compliance Deadline →