OpenAI is giving external researchers something frontier labs rarely offer: structured access to the behavioral data that could answer whether AI is actually changing work, and how. Reuters reported the program’s launch on June 8, corroborating its existence and general structure independently of OpenAI’s own announcement.
The program is led by Ronnie Chatterji, OpenAI’s Chief Economist and a former White House economic official. Academic economists Jason Furman, a former Obama administration economic adviser, and Michael Strain of the American Enterprise Institute are identified as initial research collaborators. The combination of former government officials and academic economists with different institutional backgrounds, Furman’s center-left credibility and Strain’s center-right positioning, appears deliberate. OpenAI is assembling a research advisory architecture that’s harder to characterize as ideologically predetermined.
According to OpenAI, the program will provide researchers with anonymized, privacy-governed access to usage data to study labor market effects, job design, and educational impact. The specific technical mechanism for privacy preservation, whether differential privacy, federated queries, or another approach, wasn’t specified in available source material. Don’t assume a particular method until OpenAI publishes the technical documentation.
Who This Affects
The part nobody mentions: framing the labor market impact of AI as an open research question is itself a positioning decision. OpenAI isn’t asserting what the data will show. It’s building the infrastructure to produce data that it controls access to, reviewed by economists it has selected, on a timeline it governs. That’s not a reason to dismiss the research, Furman and Strain are credible economists who’d publish uncomfortable findings, but it’s context that enterprise AI strategy teams should hold when the research comes out.
Proposal submission is reportedly open until July 5, 2026, with selection decisions expected by July 31. These dates come from OpenAI’s primary announcement document, which wasn’t accessible at time of verification. Treat them as reported deadlines requiring human validation before any researcher acts on them.
According to reporting, the program connects to a $250 million commitment through the OpenAI Foundation to worker adaptation programs. That figure requires the same caveat: it’s from the broken primary source and hasn’t been independently confirmed.
Unanswered Questions
- What technical privacy mechanism governs researcher access to OpenAI usage data, differential privacy, federated queries, or another approach?
- Will approved research findings be published publicly, or retained by researchers under confidentiality terms?
- How are research proposals evaluated, and who sits on the selection committee?
What to watch
the quality of the research that emerges. If the program produces peer-reviewed findings on AI’s labor effects that can be replicated, it will influence policy conversations at the OECD, in Congress, and in EU labor market regulation for years. If the research is selectively published or restricted, that will also be informative. Watch the first wave of approved research proposals and whether their findings are made public.
For workforce policy and compliance teams, the more immediate signal is structural: OpenAI has committed to structured external scrutiny of AI’s labor effects. Organizations using OpenAI products in workforce contexts should anticipate that this research will generate policy proposals, and probably regulatory interest, within 18 to 24 months.