As the White House and Congress remain gridlocked on a national framework for artificial intelligence, leading AI firms are taking matters into their own hands—pushing state-level legislation to shape the regulatory landscape. OpenAI and Anthropic, two of the industry's biggest players, have shifted their focus from waiting on Washington to actively crafting policy in state capitals, a strategy that is already yielding results.
Most AI companies still publicly support a unified federal standard to avoid a confusing patchwork of state laws. But with Congress moving at a glacial pace and states eager to act, firms are adapting. OpenAI's chief of global affairs, Chris Lehane, a former Clinton administration aide, told The Hill that the company is pursuing what he called "reverse federalism"—getting states to replicate each other's laws to create a de facto national framework.
OpenAI has already seen success this year, with multiple bills it endorsed passing in three major blue states: California, New York, and Illinois. The latest win came in Illinois, where state lawmakers passed SB 315, the Artificial Intelligence Safety Measures Act. The bill requires large frontier AI developers to publish and annually update safety frameworks assessing catastrophic risks, cybersecurity, and more. It also mandates third-party audits for labs with over $500 million in revenue—a provision that surprised some in the tech industry, given past opposition to similar audit requirements.
Illinois Governor JB Pritzker, a Democrat, has indicated he will sign the measure. OpenAI's endorsement of the bill was strategic, as the company sought provisions that explicitly allow future federal preemption. "This language makes it a little bit easier to ultimately, down the road, be able to put some of the political pieces in play to actually get that federal safety standard right," Lehane explained.
The company also backed New York's Responsible AI Safety and Education (RAISE) Act, signed last year, which requires AI developers to disclose safety protocols and incidents. In California, OpenAI initially had concerns about SB 53 but later expressed support after productive talks with lawmakers. The California law, signed by Governor Gavin Newsom last fall, mandates safety frameworks, transparency, and incident reporting for large AI labs.
Anthropic, the creator of the Claude AI model, has pursued a similar but distinct strategy. The firm has been engaged on state bills since 2024, well ahead of most competitors. While OpenAI has focused on blue states, Lehane acknowledged that Republican-led states have shown less appetite for such regulation. "Democratic states tend to be historically much more active in the regulatory space," he said.
The flurry of state activity comes as Americans' fears about AI's impact on jobs, the environment, and national security grow. A recent poll found that a majority now worry AI will cost them their jobs, fueling legislative urgency. Meanwhile, the Trump administration has spent over a year pushing Congress to codify federal preemption of state AI laws, but mixed feedback on the House's latest proposal and separate Senate negotiations suggest lawmakers remain far from a deal—with less than six months before a new Congress convenes.
Lehane acknowledged the ideal scenario: "In a perfect world where we could wave a magic policy/political wand, you would have legislation passed at the federal level that would establish required safety standards. As part of that, you have some type of very narrow preemption as it relates to those catastrophic-type safety risks." But until then, OpenAI and Anthropic are building the framework state by state, hoping their patchwork becomes the national norm.
