On August 4, 2026, the US Department of Justice’s Civil Rights Division announced a settlement with OpenAI and its subsidiary Statsig: the companies will pay a combined $3.2 million to resolve allegations that their green-card recruitment favored temporary visa holders and shut out US workers. The settlement, announced as press release 26-884, covers OpenAI OpCo LLC and Statsig Inc., a Bellevue, Washington-based product-development software company. It is the thirteenth settlement since the DOJ relaunched its Protecting US Workers Initiative in 2025 — a list previously made up mostly of tech companies, and now it includes the most prominent AI lab.
What the DOJ Says OpenAI Did
The dispute centers on how fewer than ten positions were recruited. According to the department, OpenAI and Statsig violated the anti-discrimination provisions of the Immigration and Nationality Act while running recruitment for Permanent Labor Certification, or PERM: those openings were not posted on the company’s external careers site like other jobs; applicants were told to mail paper applications instead of applying online; and some roles were advertised on the radio late at night — conduct the DOJ treats as deliberate discouragement of US applicants. Assistant Attorney General Harmeet K. Dhillon put it bluntly: “It is illegal to discriminate against U.S. workers by preferring temporary visa holders for jobs.”
PERM: The Paper Trail That Tripped OpenAI
PERM is the standard US process for an employer to sponsor a foreign worker’s green card, and it starts with a simple obligation: test the labor market first. The employer must advertise the job in ways the Department of Labor prescribes, and can only proceed if no able, willing, and qualified US worker applies. That makes the recruitment itself a legally scrutinized artifact — where the job is posted, how applications come in, who sees the ads. OpenAI’s problem was not that it hired foreign talent; it was that the pipeline built for that talent looked visibly different from ordinary hiring, which the regulator read as systematic exclusion.
How the $3.2 Million Breaks Down
The money splits two ways: $1.2 million is a civil penalty paid to the United States, and $2 million funds back pay for affected workers. The DOJ said the amount reflects the harm to US workers shut out of lucrative tech jobs. The settlement also imposes a compliance package: PERM openings must be posted on the public careers site and accept electronic applications, personnel must be trained on the INA’s anti-discrimination rules, employment policies must be revised, and the company submits to departmental monitoring and reporting.
OpenAI’s Response
OpenAI denied wrongdoing. A spokesperson said: “While we disagree with the DOJ’s findings, we reached this agreement to resolve the matter and move forward with our PERM program,” adding that the company must recruit globally so artificial general intelligence can “benefit all of humanity.” The statement captures a structural tension in the AI industry: frontier labs say they need the whole world’s talent, and their hiring processes are now colliding with an America-first enforcement agenda.
Settlement No. 13: Enforcement as Routine
The Protecting US Workers Initiative was relaunched in 2025 and has now produced thirteen settlements, mostly against tech companies, which is why the arrival of the best-known AI lab on that list drew attention. The backdrop is a tightening immigration posture: the Trump administration’s $100,000 fee on new H-1B visas is blocked pending litigation, but the direction is clear. For AI companies hiring at speed, every detail of PERM recruitment — where the posting lives, how applications arrive, when ads run — is now an audit target, and regulators have shown they can turn flaws in a handful of positions into a seven-figure bill.
What It Means for Tech Hiring
Three takeaways. First, PERM recruitment must be treated exactly like ordinary hiring: same careers site, same online applications — any separate channel reads as evidence of discrimination. Second, there is now an explicit list of discouragement tactics: paper applications mailed in, late-night radio ads. Details once dismissed as administrative sloppiness are named findings in a settlement. Third, the real cost is not the fine: training, policy revisions, and government monitoring run for years, and for companies headed toward the public markets, the record matters more than the amount.
Sources
- Civil Rights Division Secures Settlement with OpenAI — U.S. Department of Justice
- OpenAI settles claims of discrimination against US workers for $3.2 million — Yahoo Finance
- OpenAI pays $3.2m to settle claims it discriminated against US workers — The Guardian
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.