OpenAI opens its data pipes to economists with new research exchange

OpenAI just put its money and its data where its mouth is. The company launched the OpenAI Economic Research Exchange, a structured program that will let outside researchers tap into OpenAI’s tools and datasets to study how AI is actually changing the economy — not through breathless hot takes, but through rigorous empirical work.

The application window is open now and slams shut July 5, 2026, with selected researchers notified by the end of that month. OpenAI is specifically courting people with serious chops in applied causal inference, labor economics, productivity measurement, and related fields. This isn’t a general-interest grant program. The company wants proposals that explain exactly how “carefully governed, privacy-protected use of OpenAI tools” can answer questions that traditional datasets can’t touch. That privacy language is doing a lot of work here — it’s the clearest signal that researchers might get access to some form of usage data, not just API access.

This builds on OpenAI Signals, the company’s existing effort to measure AI’s economic footprint. But the Exchange marks a shift from internal measurement to funding external, independent work. The structure — “scoped projects with defined milestones, data governance, and review processes” — sounds closer to a corporate research partnership than an academic free-for-all. That’s probably smart. Unfettered access to proprietary data is a privacy nightmare waiting to happen. But it also raises the obvious question: how independent can the research really be when OpenAI controls the data tap and reviews the milestones?

I’m genuinely curious to see what comes out of this. The research community has been hungry for real platform data on AI adoption — not survey self-reports, not anecdotes about copywriters losing gigs, but hard evidence on who’s using these tools and with what effect. If OpenAI follows through on the “credible, independent evidence” promise, this could produce the most honest accounting yet of what generative AI is doing to labor markets. The July 2026 deadline gives researchers roughly a year to pull together proposals, which feels tight for anything truly ambitious. But for a company that moves as fast as OpenAI, maybe that’s the point.