OpenAI Economic Research Exchange: What It Funds, What It Offers, and What It Asks of Researchers

Measuring what AI is doing to employment, productivity, and economic opportunity is harder than it looks. Survey data captures stated behaviour. Administrative records measure outcomes after the fact. The behavioural signals from the tools themselves — what people actually use AI for at work, at what scale, with what effect — have remained largely inside the companies that own the platforms. On June 8, 2026, OpenAI moved to change part of that.

The OpenAI Economic Research Exchange is a new programme inviting external researchers to conduct empirical studies on AI's economic effects using approved, privacy-protected OpenAI product and usage data. Applications are open until July 5, 2026. Proposal decisions will be communicated by July 31, 2026.

What selected researchers receive

Each funded project receives a one-time grant of $25,000 for the principal investigator, a monthly research assistant stipend of $7,500, and access to approved product and usage data under a non-disclosure agreement, subject to data governance, privacy, legal, and security requirements. OpenAI is explicit on one boundary: conversation data will not be shared under any circumstances. Hands-on analysis must be performed by the visiting researcher or their research assistant, not by the OpenAI Economic Research team.

Researchers retain independence in study design and analysis. External publication is coordinated with OpenAI to cover accuracy, privacy protection, and legal compliance. The Exchange builds on OpenAI Signals, the company's existing hub for economic research and data on AI adoption, and is designed to extend the external research it can support by formalising project-based collaborations.

The ten research priorities

The request for proposals specifies ten priority areas. Labour market effects come first: the programme is seeking research that moves beyond AI exposure metrics to measure which occupations are actually seeing augmentation, displacement, new task creation, or employment growth as model capabilities develop. The remaining nine cover employer behaviour and job design, household welfare outside formal work, AI in education, unequal access and benefit, small businesses and independent work, public-sector and civic contexts, knowledge production and innovation, market structure, and how to measure AI's economic value.

The unequal access priority is framed precisely: who is not benefiting from AI, and why, with a focus on which specific barriers matter most. That framing connects directly to the financial access questions TBM has covered — where the gap between those with the skills and tools to work alongside AI and those without them is already producing measurable divergence in labour market outcomes.

Project timelines and what counts as output

The programme supports two project horizons. Short-term projects run two to six months, using existing approved data or public data to generate early findings. Medium-term projects run six to twelve months, tracking outcomes over time or leveraging quasi-experimental variation for causal claims. OpenAI states it is intentionally building a portfolio across both timelines rather than concentrating funding in longer studies.

Funded researchers must define milestones, share interim findings with the OpenAI Economic Research team, and produce at least one written output — a working paper, public brief, memo, benchmark, or dataset documentation. The minimum bar is a memo. The preference is for work that creates what OpenAI describes as tight feedback loops between evidence, internal learning, and the broader research community.

Proposals are prioritized for researchers with a publication track record in their proposed area and for those who bring unique external data that could enable more ambitious collaborations alongside OpenAI's own. The Exchange is not structured as an open grant programme — it is a set of scoped collaborations, and the RFP is clear that proposals need to specify how OpenAI tool-use data provides a distinctive empirical angle that a study could not achieve with traditional datasets alone.

Why the questions matter for fintech

The household welfare and small business priorities are particularly relevant to fintech's actual scope. The household welfare question asks what AI is doing to financial planning, decision-making, and everyday economic wellbeing outside formal employment — an area where product claims have substantially outrun independent evidence. The small business priority covers freelancers, sole proprietors, and creators, a group where AI's practical effects on work are visible in practice but poorly documented in research.

The first Exchange outputs, if the shortest project timelines hold, could enter the public domain before the end of 2026. The quality and independence of what gets published will determine whether the programme produces the kind of external evidence base it describes as its goal.

Editor's note

Every piece published on The Bright Minded goes through careful verification, but mistakes can happen. If you spot an error, have additional information, or want to flag anything, write to rosalia@thebrightminded.com.