The United States government is fracturing over how to handle the sudden explosion of artificial intelligence capabilities. Late last month, President Donald Trump abruptly postponed a highly anticipated executive order that would have forced developers to submit their frontier AI models for federal safety reviews before releasing them to the public.
The sudden cancellation highlights a fierce internal conflict between national security officials and deregulation advocates heavily backed by Silicon Valley. Tech-funded Super PACs successfully lobbied the administration to kill the safety review mandate, arguing that any federal friction would cause the U.S. to lose the global AI race to China.
These frontier models are astonishing pieces of software, capable of reasoning, coding, and solving complex problems at speeds we could barely imagine a few years ago. But that sheer computing power has national security experts sounding the alarm over potential cyber vulnerabilities. Despite these warnings, deregulation proponents won this round, convincing the President to abruptly back out of the planned safety framework. The pressure campaign included significant financial muscle from groups like the $125 million “Leading the Future” Super PAC, which boasts backing from major tech executives like OpenAI’s Greg Brockman.
Without a unified strategy from the Oval Office, individual government sectors are acting on their own. The Department of Defense is not waiting for a centralized regulatory framework. Military officials just signed agreements to deploy commercial AI systems directly onto classified military networks, eager to harness these tools for immediate tactical advantages. Meanwhile, states like California and Colorado are rushing to fill the federal void, passing their own local AI compliance laws and ignoring the administration’s stalled plans for a preemptive national standard.
What the Federal AI Regulatory Vacuum Means for State Compliance
The administration’s sudden pivot marks a complete historical reversal from the cautious oversight frameworks established by the previous administration’s 2022 AI Bill of Rights and Executive Order 14110. By scrapping the federal safety review, the White House effectively gave individual states the green light to dictate the future of tech compliance.
Software developers now face a fragmented map where deploying a single AI model requires navigating separate privacy and safety rules in California, Colorado, and potentially dozens of other states. This regulatory splintering forces startups to spend heavily on local compliance lawyers rather than pure engineering, exactly the kind of friction the tech industry hoped to avoid when they lobbied against the federal order.