OpenAI's 'Reverse Federalism' Pitch: Let States Run AI Policy First.

OpenAI has laid out a framework for AI governance that flips the usual federal-to-state policy flow on its head. In a new policy paper, the company argues for “reverse federalism” — letting state laws build the foundation for eventual national AI safety standards rather than waiting for federal action.

The proposal comes as states like California push ahead with their own AI regulations while federal lawmakers remain gridlocked on comprehensive AI policy. OpenAI suggests this patchwork approach could actually be a feature, not a bug.

State Labs for National Policy

Under OpenAI’s reverse federalism model, individual states would serve as testing grounds for different AI safety approaches. California might focus on algorithmic auditing requirements, while Texas emphasizes liability frameworks and New York tackles data privacy in AI systems.

The company argues these varied approaches would generate real-world data on what works — and what doesn’t — before scaling successful policies to the national level. Rather than one-size-fits-all federal rules written in a policy vacuum, the approach would build evidence-based national standards from proven state models.

Democratic AI Development

The proposal emphasizes “democratic AI” — systems developed with public input and accountability mechanisms built in from the start. OpenAI suggests state-level governance could make this more achievable by keeping oversight closer to affected communities.

State regulators would have more direct relationships with local businesses, universities, and civic groups using AI systems. This proximity could enable more responsive policy adjustments as AI capabilities evolve, rather than the slower federal rulemaking process.

Federal Coordination Role

OpenAI isn’t advocating for pure state-by-state chaos. The company envisions federal agencies coordinating between states, sharing best practices, and eventually harmonizing successful approaches into national frameworks.

The federal role would shift from writing initial AI rules to synthesizing lessons learned from state experiments. This could prevent a true patchwork where AI companies face conflicting requirements across state lines while still allowing policy innovation at the state level.

Industry Standards Bridge

The proposal also suggests industry standards bodies could help bridge state and federal efforts. Technical standards for AI safety testing, for example, could provide consistency even as states experiment with different regulatory approaches.

OpenAI points to how cybersecurity frameworks evolved through a mix of state laws, federal guidelines, and industry standards. The company sees a similar path for AI governance — multiple layers working together rather than top-down federal mandates.

Bottom Line

OpenAI’s reverse federalism pitch is essentially an argument for patience with the current messy state of AI policy. Instead of waiting for Congress to act, the company wants to treat state-level AI laws as valuable experiments that could inform better national policy later. Whether this leads to coherent governance or just gives AI companies more time to operate in a light-touch regulatory environment remains to be seen. The real test will be whether states actually coordinate their approaches or if reverse federalism becomes a euphemism for regulatory fragmentation.