AINEWS
The DOJ's $3.2M OpenAI settlement looks like a fine; the real lever is three years of mandatory oversight over a labor test loophole Facebook and Apple already learned to game.
The Department of Justice's $3.2 million settlement with OpenAI on Wednesday looks like a fine. It isn't one. The actual mechanism is a three-year leash: the DOJ will approve OpenAI's draft hiring policies for green-card sponsorships and receive semiannual reports on how the company recruits and screens U.S. workers. For a company valued in the hundreds of billions, $3.2 million is rounding-error money; the oversight is what does the work.
The settlement, announced by the DOJ's Civil Rights Division, ends an investigation into OpenAI's and its former subsidiary Statsig's use of the PERM labor certification process. PERM is the federal program that lets U.S. employers sponsor foreign workers for permanent residence, on the condition that the employer first test the U.S. labor market for the role. The legal bargain is built around that test. The DOJ's allegations describe a hiring process designed around it on paper and around it in practice.
According to the settlement details, OpenAI and Statsig allegedly failed to list PERM-covered roles on public job boards, ran recruitment ads in the middle of the night on local radio, and accepted paper-only applications that effectively filtered out anyone who wasn't already in the room. The settlement totals $3.2 million: a $1.2 million civil penalty plus $2 million set aside as restitution to U.S. citizens who applied to the affected roles. Fewer than ten roles were at issue, and the alleged conduct predates OpenAI's September 2025 acquisition of Statsig and its partial May 2026 divestiture. The companies did not admit wrongdoing.
The settlement is narrower than its announcement suggests. The DOJ cited five PERM cases at OpenAI between 2023 and 2025 and one at Statsig, according to the reporting. Framing the action as a crackdown misreads the scale. The civil rights division has bigger targets in mind. Its enforcement record over the past several years includes PERM settlements with Facebook in 2020 and Apple in 2021, both under the Biden administration, both describing alleged violations as "widespread and systematic" inside employer recruitment pipelines that had been quietly gamed for years.
The pattern, not the party, is the story. The Immigration and Nationality Act's PERM process dates to 1952, and the labor-test mechanism it depends on has not been meaningfully updated since. When a loophole lets employers satisfy the legal test without actually competing for U.S. workers, the system that decides who gets a green card stops functioning as a labor market and starts functioning as a sponsorship queue. Three Big Tech cases over six years, two administrations, points to the mechanism as the problem, not the politics.
The settlement's structural features are the part worth watching. The DOJ has to approve OpenAI's draft PERM hiring policies, and the company has to file semiannual reports covering foreign-employee applications, U.S.-citizen interviews, and other PERM statistics, for three years. That is the lever the civil rights division has learned to pull when the fine would not register, and it is the same lever it pulled on Facebook and Apple. Whether the oversight changes OpenAI's behavior at scale is the open question; whether the same enforcement template can be deployed against the rest of Big Tech's PERM pipelines is the longer one.
The case isn't a "gotcha" for OpenAI, and the $3.2 million isn't a meaningful financial penalty. It is a signal that the agency policing green-card sponsorships has stopped treating the labor market test as paperwork and started treating it as the regulatory perimeter it was written to be. The next three years of reports will tell whether that perimeter holds.