OpenAI Funds 14 Think-Tank Projects to Shape Policy for the 'Intelligence Age'

OpenAI is moving from writing policy white papers to bankrolling the people who write them. On August 17, the company announced the winners of its call for grant proposals to study the economic and societal impact of artificial intelligence and to develop concrete responses to the disruptions it will cause. Fourteen projects will collectively receive $1 million in direct funding, plus up to $1 million in credits for OpenAI’s models, according to details shared first with Semafor.

The scale is modest by OpenAI standards — the company raised $100 billion in 2026 at an $850 billion valuation — but the signal is not. Chris Lehane, OpenAI’s chief global affairs officer, framed the grants as an attempt to seed a policy ecosystem as ambitious as the technology itself. “To democratize the benefits of the Intelligence Age, we need policy ideas as ambitious and transformative as the technology itself,” Lehane said. “Democracies need to decide how AI is deployed, and organizations like these can serve as laboratories of democracy that develop new ways to ensure AI’s gains reach the many, not the few.”

Where the money comes from

The grants trace back to OpenAI’s April 2026 policy document, “Industrial Policy for the Intelligence Age,” which proposed everything from new benefit systems to national wealth-building vehicles for an AI-driven economy. The paper argued that broad access to AI tools — ChatGPT now claims roughly a billion users — is only half the equation. Whether the technology’s gains are broadly shared depends on policy choices made now, not incremental adjustments later.

Rather than leaving those ideas as internal advocacy, OpenAI opened a competitive call for proposals earlier this year. More than 400 submissions arrived. The 14 winners announced this week span the political spectrum in the United States and stretch across four continents — Europe, South America, and Asia alongside the American institutions.

The winners and what they’ll build

Workforce and economic opportunity. The American Enterprise Institute, a conservative think tank, will team up with the liberal Urban Institute on a joint project modeling AI’s disruption of the American workforce — a rare explicitly cross-ideological pairing. The Progressive Policy Institute is designing a “person-based benefits system” that would attach social protections to individuals rather than jobs, an idea aimed directly at the decoupling of work from employment benefits. The Tax Foundation will analyze what AI-driven automation does to tax revenue, and the Windfall Trust will convene national working groups on AI economic policy.

International and rights-based approaches. The Centre for European Policy Studies will examine how AI-driven productivity gains are distributed across Europe, while the European Centre for International Political Economy is developing a framework for a “Right to AI” — exploring ownership and access models that go beyond traditional employment relationships. In Brazil, IMPA (the Institute for Pure and Applied Mathematics) will study what research institutions need to integrate AI, and HCFMUSP will evaluate AI-enabled clinical infrastructure for public hospitals — one of the few winners focused on frontline public-service delivery.

Safety and resilience. A significant block of grants goes to governance and national security work. The Institute for Security and Technology is building a framework for measuring and responding to recursively self-improving AI systems — systems that improve their own capabilities, a frontier that current regulation barely touches. The Nuclear Threat Initiative, better known for nonproliferation work, will explore secure cross-border information sharing for AI-driven biological risks. The Council on Strategic Risks aims to build frontier-AI-safety capacity among national security professionals. Nanyang Technological University in Singapore is developing privacy-preserving AI agents for auditing government policy simulations, and Yonsei University in South Korea will assess legislative oversight quality to study AI-enabled democratic accountability.

The Abundance Institute rounds out the list with work on the energy infrastructure needed to support data center demand — a reminder that AI policy is increasingly energy policy.

Why it matters

Three things make this announcement more interesting than a routine corporate philanthropy exercise.

First, the ideological spread is deliberate. Funding AEI and the Progressive Policy Institute in the same round is a hedge against the accusation that OpenAI is buying friendly research. The “laboratories of democracy” framing borrows from American federalism — the idea that states (here, think tanks) can experiment in parallel and the best approaches win.

Second, it’s self-interested in an unusually transparent way. Several projects will effectively stress-test OpenAI’s own April policy proposals — the company explicitly describes a commitment to “testing and building upon” its own ideas. If a person-based benefits system or a Right to AI framework survives contact with independent researchers, OpenAI gains evidence to bring to legislators; if it doesn’t, the company learns that cheaply.

Third, the timing. The grants land amid a live regulatory fight: the EU is in talks with OpenAI and Anthropic over AI Act enforcement following recent cyber incidents involving their models, and the White House convened the major AI labs in early August to finalize voluntary safety-testing commitments. Seeding a network of policy institutions that share OpenAI’s vocabulary — economic opportunity, societal resilience, Intelligence Age — is a soft-power play that complements the harder lobbying tracks.

What to watch

The projects run for six months, with results expected in 2027. The interesting question isn’t whether the research is good — these are credible institutions — but whether any of it changes what legislatures actually do. A $2 million program (cash plus credits) won’t move macro policy on its own. But if even a few frameworks — a workable benefits design, a biorisk information-sharing protocol, a measurable definition of self-improving systems — get picked up by governments, OpenAI will have bought itself a seat at the table where the Intelligence Age’s rules get written. And at $2 million, that’s the cheapest policy investment in the company’s history.