US Advances AI Safety Frameworks in 3 States as Trump Team Targets August Federal Testing Rules
Updated
Updated · OpenAI · Jul 15
US Advances AI Safety Frameworks in 3 States as Trump Team Targets August Federal Testing Rules
3 articles · Updated · OpenAI · Jul 15
Summary
California, New York and Illinois have aligned on frontier AI safety laws, creating a de facto national baseline while the Trump administration drafts federal testing rules for the most capable models.
Early August is the administration’s target for a cyber-testing framework that would set standards, timelines and processes for evaluating advanced AI systems before wider government and critical-infrastructure use.
OpenAI says the state laws converge on three core requirements: documented safety frameworks with public risk disclosures, reporting of serious incidents, and independent audits.
Congress is also weighing bipartisan federal proposals, with OpenAI arguing Washington—not states—should lead national-security reviews and strengthen the Center for AI Standards and Innovation.
The broader aim is a single US standard that avoids a state-by-state patchwork and can anchor a US-led international AI governance framework discussed at the G7 and beyond.
Can a national AI standard prevent a complex web of state regulations from stifling American innovation?
Are current security measures enough to stop rivals from copying US AI and erasing America's technological lead?
As AI agents deploy faster than safety rules, what unseen societal risks are being overlooked in the rush?
The 2026 U.S. AI Regulation Report: Federal Initiatives, State Laws, and the Battle for Preemption
Overview
The report explores how the Trump administration is shaping federal AI regulation by balancing innovation with national security and safety concerns. After President Trump signed a major executive order in June 2026, the federal government signaled a shift toward more oversight, recognizing that AI is rapidly changing American life and raising urgent questions about jobs, privacy, and accountability. The new framework is currently voluntary but encourages AI developers to work closely with national security agencies. This approach aims to create a robust system that can adapt to future political changes, setting the stage for possible stronger federal regulation as AI continues to evolve.