OpenAI Launches an Economic Research Exchange to Fund Outside Study of AI's Impact

OpenAI Launches an Economic Research Exchange to Fund Outside Study of AI's Impact

The company that did more than any other to set off the generative-AI boom now wants academic economists to grade its homework. On June 8, OpenAI announced the OpenAI Economic Research Exchange, a grant program and data-sharing platform designed to bankroll independent, peer-quality research into how artificial intelligence is reshaping jobs, productivity, firms, and the broader economy.

The pitch is straightforward, and notably modest in tone for a company valued in the hundreds of billions: the questions surrounding AI and the economy are too important to be answered by anecdote, and OpenAI is offering money plus access to its proprietary usage data to researchers willing to study them rigorously. "Understanding those changes will require more than anecdotes," the company wrote in its launch announcement. "It will require rigorous empirical research, grounded in real-world evidence and pursued by a broad community of researchers."

Applications opened immediately and close July 5, 2026, with selected researchers to be notified by July 31, according to OpenAI's request for proposals.

What the Exchange actually offers

Under the program, OpenAI says it will fund "structured, project-based collaborations" with its in-house OpenAI Economic Research team. Selected investigators retain independence over study design and analysis — a point the company stresses repeatedly — while gaining something academics rarely get: access to approved, privacy-protected OpenAI product and usage data that "would extend beyond traditional datasets alone."

The financial terms, detailed in the request for proposals, are concrete. Each selected project receives a one-time $25,000 research grant for the principal investigator or investigators, plus $7,500 per month to cover a research-assistant stipend or contractor compensation. Projects are expected to run on defined milestones with formal data-governance and review processes attached.

OpenAI says it is casting a wide net for talent, welcoming proposals from researchers with "strong empirical skills" across applied causal inference, measurement, labor economics, productivity, education, entrepreneurship, public finance, regional economics, development, and inequality. Submissions will be judged on methodological rigor, feasibility, fit with the Exchange's priorities, clear milestones, and the potential to produce credible external evidence.

The Exchange does not arrive in a vacuum. OpenAI frames it as an extension of OpenAI Signals, its internal effort to measure AI's economic footprint, and the program sits alongside a far larger commitment announced weeks earlier: in late May, the OpenAI Foundation — the company's nonprofit arm — pledged an initial $250 million toward AI's economic impact, earmarked for labor-market measurement, worker support, and new mechanisms for distributing AI-generated wealth. The Exchange is, in effect, the academic-research wing of a much broader policy push. The effort is led internally by OpenAI's first chief economist, Aaron "Ronnie" Chatterji, a Duke University professor who previously served as the Biden administration's CHIPS Act coordinator.

Why a frontier lab funding economic research matters

For policymakers, the appeal is obvious. The single biggest gap in the AI-and-jobs debate is data. Labor-market statistics are slow, lagging, and ill-suited to capturing how a tool like ChatGPT changes what a worker does inside an existing job rather than whether that job exists at all. OpenAI sits on exactly the granular usage data — what people ask AI to do, in which occupations, at what scale — that economists have been unable to obtain. Opening even a privacy-screened slice of it to outside researchers could meaningfully sharpen an empirical picture that has so far leaned heavily on surveys and projections.

That is also precisely what makes the move worth scrutinizing. A frontier lab funding the research that will shape public understanding — and eventually regulation — of its own product raises unavoidable questions about independence and conflict of interest. OpenAI has commercial and political stakes in how the AI-jobs story is told: a narrative of broad productivity gains and new categories of work is far friendlier to its business and its policy agenda than one of mass displacement.

OpenAI's answer is procedural. It promises researchers will "retain independence in study design and analysis" and emphasizes governed, privacy-safe data handling. But independence in design is not the same as independence in agenda-setting. The company chooses which proposals to fund, controls which data is approved for release, and defines the "Exchange priorities" against which projects are evaluated. Researchers may freely analyze the data they are given — but the gatekeeping over what data exists, and which questions get money, remains with OpenAI. Academic norms around disclosure of funding will matter enormously to how this work is received; a $25,000 grant and proprietary-data access is a relationship that any credible journal or referee will want declared.

There is precedent worth remembering here, and it is not entirely reassuring. Tech-funded academic research — from social-media platforms to the gig economy — has a mixed record, with recurring fights over data access being throttled the moment findings turn unflattering. The structure OpenAI has built, with the company holding the data keys, is the kind of arrangement that works smoothly until a researcher produces something the funder dislikes.

The debate the data is supposed to settle

All of this lands in the middle of an unresolved argument. One camp warns that AI is poised to hollow out entry-level white-collar work — software, customer service, analysis — faster than new roles can replace it. Another holds that AI is more likely to augment workers and lift productivity, reshaping industries rather than emptying them, much as past general-purpose technologies eventually did. The honest answer at this stage is that nobody knows, because the rigorous, real-time measurement does not yet exist. That gap is exactly the one the Exchange is meant to fill — and exactly why who fills it, and on whose terms, is not a neutral question.

It is also part of a broader industry pivot. As Republic World noted in its coverage, the AI conversation is "increasingly moving beyond model performance and product launches towards long-term economic and societal consequences." OpenAI is hardly alone; rival Anthropic has run its own economic-index work tracking AI usage by task. But OpenAI's combination of a $250 million foundation commitment, a chief economist drawn from the policy world, and now a structured external-research exchange amounts to the most deliberate attempt yet by a frontier lab to shape — and be seen shaping — the economic evidence base.

What to watch

The first real test comes after July 31, when OpenAI names its inaugural cohort. Watch for the calibre and independence of the chosen researchers and institutions, the breadth of the data actually made available, and — crucially — whether any early findings cut against OpenAI's commercial interests and still see daylight. Whether the Exchange becomes a genuine engine of independent evidence or a well-funded exercise in narrative management will be judged not by its launch announcement but by the first study it funds that OpenAI would rather not read.