OpenAI's Chris Lehane outlines a "reverse federalism" approach where state laws build a national framework for safe, democratic AI. California, New York, and Illinois have advanced frontier safety legislation establishing a common baseline for governing powerful AI systems. These states align on core elements: documented safety frameworks with risk assessments and public disclosure, serious incident reporting, and independent audits. This convergence creates a de facto national standard, avoiding a patchwork that could slow innovation and undermine a global democratic AI stack.

At the federal level, the Trump Administration is developing a testing framework for the most capable models on cybersecurity, targeting early August completion. OpenAI supports strengthening the Center for AI Standards and Innovation (CAISI) to lead federal evaluation of advanced systems. Congress is also advancing bipartisan proposals, notably from Reps. Jay Obernolte and Lori Trahan, to codify a national frontier safety regime. OpenAI argues federal leadership on testing, clear company requirements, and state-federal reinforcement are essential. A coherent system ensures the best testers evaluate the most capable models and trusted defenders gain rapid access to stay ahead of malicious actors.