How IER Described the OpenAI and Statsig Pattern
For OpenAI and Statsig, the settlement is the clearest available record of what the Justice Department’s Immigrant and Employee Rights Section (IER) investigated and what the parties agreed to do. It is not, however, a complete case file or a court judgment. The task here is to reconstruct the two investigation tracks, compare the scope attributed to each company, and use the settlement’s own procedural language so that the reader can distinguish an agency finding, a denial, a negotiated obligation, and an unresolved fact.
Two investigations, one described recruitment mechanism
The agreement says IER initiated an independent investigation of OpenAI under 8 U.S.C. § 1324b(d)(1) on August 13, 2025. The stated purpose was to determine whether OpenAI had engaged in citizenship-status discrimination in recruitment and hiring related to PERM, in violation of 8 U.S.C. § 1324b. On the same date, IER also initiated three charge-based investigations of OpenAI related to PERM recruitment. The agreement groups these investigations together as the OpenAI Investigations.
The labels tell us how the agreement organizes the inquiry, but they do not give us the underlying record. The OpenAI investigation is called independent and is placed under the cited statutory provision. The other three are called charge-based. The settlement does not reproduce the charges, identify the people who brought them, or explain how the investigative steps differed. We can therefore distinguish the two procedural labels without claiming that one label proves a different outcome, a different evidentiary standard, or a different view of the facts. What the agreement does establish is that several OpenAI-related tracks were resolved together rather than being treated as one unexplained complaint.
IER initiated a separate charge-based investigation of Statsig on September 9, 2025. The agreement calls that investigation, together with the OpenAI Investigations, “the Investigations.” Statsig is therefore a separate subject in the settlement, not merely an example folded into the OpenAI account. The common subject is PERM-related recruitment; the investigative path and the scope attributed to each company are not identical. The dates are useful for chronology, but they do not by themselves show when a particular advertisement was created, when a worker encountered it, or when a company changed its process.
The agreement’s background then states what the Investigations found. For OpenAI, IER found reasonable cause to believe that, beginning no later than June 2023 and continuing until at least April 2025, the company engaged in a pattern or practice of citizenship-status discrimination in five PERM-related advertisements, in violation of 8 U.S.C. § 1324b(a)(1)(B). For Statsig, IER determined that the company violated the same cited provision in one PERM-related recruitment in July 2025. The wording gives OpenAI a broader reported period and a pattern-or-practice characterization, while the Statsig account is tied to one recruitment in a specified month.
These are not interchangeable descriptions. A finding about five advertisements over a period extending from no later than June 2023 to at least April 2025 has a different stated scope from a finding about one recruitment in July 2025. The difference matters even when the same general mechanism is alleged. It affects what a careful reader can say about recurrence, duration, and organizational reach. It does not allow the reader to infer that the five OpenAI advertisements affected the same number of people, involved the same job titles, or operated through identical internal processes.