New policy ideas for the Intelligence Age

OpenAI is funding 14 independent organizations to develop and test concrete AI policy ideas, distributing $1 million in direct grants plus up to $1 million in API credits across projects in the United States, the European Union, Brazil, Singapore, and South Korea. The 14 recipients were selected from more than 400 proposals.

Details

• Origin: The program follows OpenAI’s April 2026 paper “Industrial Policy for the Intelligence Age,” and is framed around the idea that “whether AI ultimately serves the many rather than the few will also depend on the choices societies make about how to deploy it”
• Structure: Each project runs for six months, with results expected in 2027. OpenAI says the goal is to turn high-level policy concepts into practical, testable implementations that governments can adapt locally
• Economic opportunity track:
  • American Enterprise Institute — an AEI-Urban bipartisan commission modeling AI-driven workforce disruption scenarios and policy responses
  • Centre for European Policy Studies — examining how AI-driven productivity gains are distributed across Europe and developing social investment models
  • European Centre for International Political Economy — a “Right to AI” framework and models for productive asset ownership
  • Abundance Institute — state-level energy policy work tied to data center infrastructure expansion
  • Progressive Policy Institute — prototypes for a person-based benefits system and livelihood insurance
  • Tax Foundation — analysis of AI’s impact on tax revenues and related policy recommendations
  • Windfall Trust — international working groups on fiscal, labor, and competitiveness questions
  • Instituto de Matemática Pura e Aplicada (Brazil) — research infrastructure for AI-enabled scientific advancement
  • Hospital das Clínicas (Brazil) — AI-enabled clinical infrastructure for public health systems
• Societal resilience track:
  • Institute for Security and Technology — a framework for measuring and coordinating responses to recursively self-improving AI systems
  • Nuclear Threat Initiative — international information-sharing protocols covering AI and biological risks
  • Council on Strategic Risks — AI-safety training for national security professionals
  • Nanyang Technological University — privacy-preserving AI agents for policy simulation
  • Yonsei University (South Korea) — AI-based methods for assessing the quality of legislative oversight
• Stated principle: OpenAI explicitly says “these choices should not be made by technology companies alone,” positioning the grants as a way to seed independent, government-adaptable research rather than OpenAI-authored policy

What happened next

This is a funding and prototyping initiative rather than a product launch — the actual policy outputs aren’t due until 2027, when each six-month project wraps. It extends a pattern OpenAI has followed with prior initiatives like the OpenAI Economic Research Exchange and its research partnership with the APA: fund external, independent researchers rather than publish OpenAI’s own policy recommendations directly. Whether any of the 14 prototypes get adopted by the governments they target will be the real test of the program.