OpenAI Endorses 'Reverse Federalism' Model for US AI Safety Standards
OpenAI's chief global affairs officer argues state-level legislation in California, New York, and Illinois is creating a de facto national AI safety framework.
Last verified:
State-Level AI Safety Legislation Creating Convergence Path
OpenAI Chief Global Affairs Officer Chris Lehane argues that a patchwork of state-level AI safety laws—rather than gridlock in Congress—is establishing the foundations for a coherent national frontier AI governance framework. According to the OpenAI Blog, California, New York, and Illinois have each advanced legislation that mandates safety assessments, incident reporting, and independent audits of frontier models, creating what Lehane frames as a “reverse federalism” model in which states lead standard-setting.
Lehane’s argument reframes a regulatory risk that typically concerns industry: rather than viewing state-level divergence as fragmentation, OpenAI positions aligned state laws as a preferable alternative to either federal preemption or chaotic, misaligned rules across jurisdictions. The framing acknowledges a political reality—federal AI legislation has stalled—and converts it into a governance opportunity.
Three-Part Safety Framework Emerging Across States
According to the OpenAI Blog, the legislation passed or advanced in these three states converges on a common structure: documented safety frameworks with risk assessments and public disclosure; mandatory reporting of serious safety incidents; and independent, objective audits for governance and accountability.
OpenAI credits California with establishing “the core disclosure framework,” New York with demonstrating adoption could scale across jurisdictions, and Illinois with adding the requirement for “independent verification” of compliance. The characterization treats these laws not as competing rules but as complementary components of a single national baseline that emerged incrementally rather than through centralized lawmaking.
Why This Matters
The argument cuts two ways for frontier labs like OpenAI. On one hand, Lehane is advocating that states continue aligning—which protects OpenAI from the downside risk of 50 different regulatory regimes. On the other hand, he is implicitly warning federal policymakers that absent federal action, state-level regulation will continue advancing, and labs may find themselves subject to governance they did not help design.
For state policymakers, the OpenAI position suggests that further alignment is feasible and may be preferable to independent regulatory experimentation. For other nations, Lehane signals that OpenAI views this state-coordinated model as a foundation for exporting US-style democratic AI governance globally—a claim that will likely face scrutiny from other countries’ regulators, who may see this as an attempt to establish US standards as international norms rather than true multilateral governance.
The timeline of state adoption—California, then New York, then Illinois—suggests momentum, but the durability of this “reverse federalism” approach depends on whether states continue to converge or diverge as frontier AI capabilities and risks evolve.
Frequently Asked Questions
What is 'reverse federalism' in the context of AI regulation?
According to OpenAI, it is a model where state-level governments establish shared governance principles that eventually converge into a de facto national standard, rather than waiting for federal legislation.
Which states have passed frontier AI safety legislation?
California, New York, and Illinois have each advanced frontier safety laws that include safety frameworks, incident reporting requirements, and independent audit provisions.
Why does OpenAI prefer state coordination over a federal approach?
OpenAI argues that aligned state-level laws create a coherent national baseline while avoiding regulatory fragmentation that could slow innovation or undermine US competitiveness against other nations.
What are the three core elements OpenAI identifies in aligned state laws?
Documented safety frameworks with risk assessments and public disclosure, mandatory reporting of serious safety incidents, and independent audits for governance and accountability.