OpenAI puts $1 million behind 14 research projects on AI's economic fallout
OpenAI is backing independent research into how AI could reshape jobs, opportunity and safety in the intelligence age.
OpenAI announced on 17 August that it is funding 14 independent research projects examining how societies should prepare for AI's effects on jobs, taxation, social protection, energy and scientific research. The company is providing $1 million in total grants plus up to $1 million in API credits, selected from more than 400 responses to a call for proposals issued in April.
The projects span the United States, the European Union, Brazil, Singapore and South Korea. India is not among them, despite being one of ChatGPT's largest user bases and a market where the labour-market questions the programme is asking apply at scale.
What the grants follow from
The programme delivers on a commitment made in Industrial Policy for the Intelligence Age, an OpenAI paper published in April 2026 that argued incremental policy updates would not be sufficient as AI capabilities advance. OpenAI offered the paper explicitly as a starting point for discussion rather than a final set of recommendations, and invited outside groups to build on, refine or challenge it.
The company's framing is that access to AI tools alone will not determine who benefits. Whether AI serves the many or the few, it argues, depends on how societies choose to deploy it and distribute the gains.
Who got funded
On the economic side, the American Enterprise Institute will run a bipartisan commission with the Urban Institute building low, moderate and high-disruption scenarios for AI's effect on employment and connecting disruption indicators to policy playbooks. The Tax Foundation will model how AI adoption shifts the balance between labour income, corporate profits and capital gains as sources of public revenue. The Progressive Policy Institute will design a person-based benefits system covering retirement, health, leave, education, training and disability that travels with the worker rather than the job, prototyping a "livelihood insurance" mechanism triggered by broad changes in an occupation.
In Europe, the Centre for European Policy Studies will benchmark how AI productivity gains are distributed across EU workers, firms and regions against the US picture. The European Centre for International Political Economy will build a framework for a "Right to AI," comparing employee ownership, citizen investment, pension-based ownership and social wealth funds. The Abundance Institute will study how US states can expand power generation and transmission against data centre demand.
Two Brazilian projects are notable for emerging-market relevance. IMPA will study what research institutions actually need to convert AI access into scientific output, including compute, training and local infrastructure. Hospital das Clínicas at the University of São Paulo will run a pre-deployment evaluation of AI clinical infrastructure for SUS, Brazil's public health system, using synthetic and de-identified data.
The safety side
Four projects address institutional resilience. The Institute for Security and Technology will build a measurement and coordination framework for recursively self-improving AI systems, including observable indicators of uncontrolled self-modification and escalation criteria linking technical signals to coordinated response. The Nuclear Threat Initiative will test the legal feasibility of cross-border information sharing on AI and biological risk. The Council on Strategic Risks will train national-security professionals in frontier AI safety. Yonsei University will test whether AI methods can assess the quality of legislative oversight in South Korea's National Assembly.
The obvious question
A company funding independent research into policy that will govern that company invites scrutiny. OpenAI's answer is partly structural: recipients span the US political spectrum, from AEI on the right to the Progressive Policy Institute on the left, and the company says independent institutions should challenge assumptions including its own. Chris Lehane, OpenAI's chief global affairs officer, framed the programme as needing policy ideas as ambitious as the technology.
The counter-argument is that $1 million is a rounding error against OpenAI's capital position, and that funding the research agenda is itself a way of setting it.
Projects run for six months, with results due in 2027.