US Prioritizes AI Velocity Over Safety in Geopolitical Race

The De-Regulation Gambit

In a dramatic policy reversal, the Trump administration abruptly canceled a significant executive order on artificial intelligence just hours before its scheduled signing. This decision, framed by the White House as a move to prevent stifling economic growth and America's lead in the AI sector, underscores a strategic pivot: prioritizing raw velocity in AI development over the implementation of safety measures and regulatory oversight. President Trump indicated that initial considerations for oversight were deemed potential "blockers" to the surging economy and job creation, articulating a belief that any friction, even in the interest of safety, is an unacceptable risk in the current geopolitical "AI arms race."

This "blocking paradox" reflects deep-seated anxieties regarding China's accelerating technological advancements. While the U.S. maintains a lead in AI model performance, the narrowing gap has propelled American policymakers toward a binary choice: either embrace stringent safety oversight and risk falling behind, or aggressively pursue rapid innovation. By shelving the executive order, the U.S. has effectively committed to a "black box" approach to AI development, driven by the perception that bureaucratic regulation cannot counter the pace of Eastern technological progress.

The Ascent of Agentic AI and Escalating Risks

This pursuit of velocity, however, directly confronts a new architectural reality: the rise of "agentic AI." As described by Michael Dell, AI is evolving beyond a mere tool for human users into autonomous "non-human workers" operating at machine speed. This shift dramatically alters the cybersecurity landscape. Unlike traditional software, where breaches often have localized impacts, an autonomous ecosystem amplifies the "blast radius" of any compromise. A corrupted autonomous agent or one trained on malicious data can propagate across an entire organization's workflows in milliseconds, outpacing human oversight and creating unprecedented security challenges.

The "Mythos model" exemplifies this dual nature. In just over a month, this AI has identified 10,000 bugs in critical software across various sectors, demonstrating a revolutionary shift in cybersecurity labor economics. A single AI model can now execute the audit work of thousands of human analysts in a fraction of the time. Yet, this formidable capability presents a significant vulnerability: the same power that secures digital infrastructure can be weaponized if it falls into malicious hands. This creates a "wild circle," an escalating loop where increasingly powerful AI is required to defend against threats empowered by AI itself, raising concerns about a cycle of digital warfare that could rapidly exceed human intervention capacity.

Regulatory Vacuum and Industry Influence

The cancellation of the executive order leaves a substantial regulatory void. Notably, the proposed order, even before its withdrawal, was considered toothless, offering only a voluntary framework for model submissions rather than mandated legal requirements. Cybersecurity attorney Lisa Garber highlighted that even this minimal "suggestion" of safety proved too much for the industry. Reports suggest significant influence from prominent tech billionaires, who successfully lobbied against even non-binding oversight, thereby shaping policy toward deregulation.

This absence of a formal regulatory framework carries tangible human consequences. Without guardrails, the battle extends beyond algorithms to immediate societal harms, including "nudify" apps, deepfakes, and critical child safety threats. By opting for a "Wild West" approach to placate industry leaders, the administration effectively externalizes the risks of AI onto the most vulnerable segments of society.

The narrative of a singular "race" against China, while prevalent, oversimplifies a more complex reality. Garber notes that multiple races are underway concurrently: for power, talent, and creativity. The U.S. currently leads in these areas, but maintaining this dominance does not necessitate a lawless environment. The argument for guardrails is not a call for stagnation but for building a resilient foundation. Integrating expertise from cybersecurity, ethics, and law could enable the U.S. to sustain its creative edge while safeguarding its citizens. The ultimate challenge lies in balancing global leadership with domestic safety, fostering innovation without ceding control, and ensuring that the pursuit of speed does not come at the cost of foresight.