President Trump's decision to delay a proposed artificial intelligence executive order reflects the right concern: America cannot afford AI rules that weaken its competitive edge. But that does not mean doing nothing.
There is a vast difference between regulating AI into paralysis and scrutinizing the small number of frontier systems that could create real national security risks. The first would be a disaster. The second is common sense.
AI is no longer just a productivity tool. The most advanced models are becoming part of the terrain of cyber defense, intelligence, military planning, critical infrastructure, and scientific research. The same tools that help companies write code and analyze data can help hostile actors find software vulnerabilities, automate intrusions or produce disinformation at scale.
That does not make every AI tool a threat. It means the United States needs to know when a system becomes powerful enough to matter.
This is where Washington's AI debate has gone off the rails. Some critics hear "oversight" and imagine a federal takeover of the tech sector. Others want to use AI panic to regulate ordinary algorithmic tools that have little to do with national security.
Both are wrong.
A targeted review of frontier AI models is not anti-innovation. It is basic national defense. The United States already pays special attention to technologies that affect weapons systems, intelligence capabilities, advanced semiconductors, and critical infrastructure. We do not regulate every software company like a defense contractor. But we do not ignore systems that could be exploited by adversaries either.
AI deserves the same discipline.
Property management platforms, pricing software, logistics systems, fraud detection tools, and other commercial technologies are increasingly being portrayed as inherently dangerous simply because they use algorithms or automation. In reality, most of these tools are designed to improve efficiency, reduce waste, help businesses respond to market conditions, and lower operational costs — not threaten consumers or national security.
Recommended
The key is drawing the line where the real risk begins. A tool that improves customer service is not the same as a model that can materially assist cyber operations against hospitals, energy grids, financial infrastructure, or military systems.
America's adversaries understand this. China is investing heavily in advanced computing, military AI and cyber capabilities. Criminal groups are already learning how to automate fraud and intrusion. State-backed actors will not pass up tools that make attacks faster, cheaper and harder to trace.
Pretending otherwise is not pro-market. It is naïve.
Vigilance is not the same as control. Washington should not build a sprawling AI bureaucracy that sweeps in half the digital economy. It should not treat every algorithm used in housing, finance, advertising, logistics, or workplace management as a crisis requiring a response.
That would be exactly the kind of overreach critics fear and companies rightly oppose.
America's AI advantage comes from private-sector speed, world-class researchers, deep capital markets and a culture that rewards building. The government did not create that ecosystem. It can, however, damage it.
So the right approach is narrow, predictable, and capability-based, focusing on the small number of frontier systems that could pose real national security risks and leaving ordinary commercial innovation alone.
That would give companies clearer rules, investors more confidence, and the public a better assurance that Washington is watching the right threats instead of chasing headlines.
This is the balance the Trump administration should strike.
The president does not have to choose between AI leadership and national security. America needs both. It must remain the best place in the world to build AI while making sure the most consequential systems are not released blindly into a dangerous world.
The critics will say any review is the first step toward a crackdown. That is disingenuous.
The real danger is not that Washington examines AI systems that could be abused by foreign adversaries. The danger is that America talks itself out of basic security while China, cybercriminals, and hostile states move faster.
AI will help define the next era of economic and military power. The United States should not smother it. But it would be foolish to sleepwalk into it.
Reynold Schweickhardt is a non-resident senior fellow at the Foundation for American Innovation.
Editor’s Note: Do you enjoy Townhall’s conservative reporting that takes on the radical Left and woke media? Support our work so that we can continue to bring you the truth.
Join Townhall VIP and use promo code FIGHT to receive 60% off your membership.
Join the conversation as a VIP Member