OpenAI announced the OpenAI Economic Research Exchange, a program designed to support external empirical research on how artificial intelligence affects workers, firms, institutions and the broader economy. The announcement was published on June 8, 2026, and an update on August 5, 2026, noted a current research cohort.
Program goals and approach
OpenAI said the Exchange will fund structured, project-based collaborations between selected researchers and OpenAI Economic Research. The stated aim is to produce credible, independent evidence about AI’s economic impacts by combining rigorous empirical methods with carefully governed access to OpenAI tools and datasets.
OpenAI emphasized that projects should rely on real-world evidence rather than anecdotes and that research proposals must explain how privacy-protected use of OpenAI tools would help answer specific questions. Selected projects will include defined milestones, data governance arrangements and review processes, according to the announcement.
Who should apply and how projects will be evaluated
The Exchange welcomes proposals from researchers with empirical expertise and relevant domain knowledge, including applied causal inference, measurement, labor economics, productivity, firms, education, entrepreneurship, public finance, regional economics, development and inequality. OpenAI said it seeks projects that address important questions about AI’s economic effects and that would extend beyond traditional datasets.
Proposals will be evaluated on methodological rigor, feasibility, alignment with Exchange priorities, clarity of milestones and potential to contribute credible external evidence on economic impacts. OpenAI noted the Exchange builds on broader efforts such as OpenAI Signals to improve measurement of AI’s effects.
Applications are open now and will close on July 5, 2026. Submissions will be reviewed and selected researchers will be notified by July 31, 2026. For application details or questions, OpenAI provided the contact email econresearch@openai.com.
Original source: OpenAI News