We need guardrails for artificial superintelligence NOW — before it’s too late

America’s “AI race with China” is a headline we see more and more. But we’re actually in two high-stakes races with China in artificial intelligence.

First: a competition for commercial dominance that is reshaping economies, military power and global influence.

The second race, though less visible, has the potential to be even more existential: a sprint toward artificial superintelligence.

What’s ASI? Unlike current AI models trained to perform relatively narrow tasks, ASI refers to a hypothetical future version of AI that exceeds human intelligence across every domain — creative, strategic, even emotional.

It could be capable of autonomously improving itself, outpacing our ability to control or predict it.

This technology doesn’t yet exist, but leading experts, industry leaders and lawmakers believe its emergence could be possible within the next decade.

That’s the problem: It may not feel urgent — until it’s too late. Which is why the time to act is now.

President Donald Trump and his team are in a unique position to secure America’s preeminence on both fronts by winning the commercialization race and negotiating what may be the most consequential diplomatic deal since the nuclear-arms treaties of the Cold War.

China’s advancements in commercial AI have dramatically closed America’s lead on the rest of the world. How? Beijing bought, stole and downloaded US technology, leading to breakthroughs that resemble a modern-day Sputnik moment.

Chinese firms are unveiling AI models that are both cheaper and more sophisticated than we knew possible (remember our reaction to DeepSeek?).

China’s state-directed pursuit extends far beyond economic ambitions. The Chinese Communist Party openly seeks a technological dominance that’s anchored in its own core principles: surveillance, censorship, and control.

A Chinese-led AI era risks embedding these authoritarian pillars into the digital fabric of global civilization and everyday life.

An unregulated race toward ASI presents an even deeper danger. Influential forecasts — notably AI 2027, a predictive framework developed by key experts — warn that the emergence of ASI could pose unprecedented risks to humanity.

Yes, these risks are still theoretical, but they’re also not so far-fetched. In the hands of an adversary, an ASI system has the potential to destroy global electrical grids, develop incurable super-viruses or empty every bank account in the world.

That may sound like the latest plot in Tom Cruise’s “Mission Impossible,” but it’s also the plausible consequence of unchecked superintelligence in the wrong hands.

Most top AI leaders believe ASI could materialize within this decade and pose unprecedented risk.

Ilya Sutskever, co-founder and chief scientist of OpenAI, told his researchers: “We’re definitely going to build a bunker before we release [artificial general intelligence].” So, if bunkers are the recommended precaution for AGI, what should we prepare for ASI?

Vice President JD Vance appears to be grappling with these risks, as he reportedly explores the possibility of a Vatican-brokered diplomatic slowdown of the ASI race between the United States and China.

Pope Leo XIV symbolizes precisely the kind of neutral, morally credible mediator capable of convening such crucial talks — and if the Cold War could produce nuclear-arms treaties, then surely today’s AI arms race demands at least an attempt at serious discussion.

Skeptics naturally and reasonably question why China would entertain such negotiations, but Beijing has subtly acknowledged these undeniable dangers as well. Some analysts claim Xi Jinping himself is an “AI doomer” who understands the extraordinary risk.

Trump is uniquely positioned to lead here. He can draw a clear line: America will outcompete China in commercial AI, no apologies. But when it comes to ASI, the stakes are too high for brinkmanship.

We need enforceable rules, verification mechanisms, diplomatic pressure and, yes, moral clarity — before this issue gets ahead of us.

During the Cold War, Presidents Dwight Eisenhower, John Kennedy and Ronald Reagan all knew that competing militarily didn’t mean refusing to negotiate guardrails.

Reagan’s mantra — “trust but verify” — is just as relevant for ASI as it was for nuclear arms.

This is President Trump’s opportunity. He can drive the AI economy forward, infusing American founding principles into global AI adoption, while leading a parallel effort to prevent catastrophe.

Done right, this would be the most consequential diplomatic initiative since the Strategic Arms Reduction Treaty. And it wouldn’t come at the cost of American strength; it would cement it.

We’ve reached a crossroads. The commercialization of AI can secure America’s future, but the weaponization of superintelligence could end it.

Chris Stewart was a member of Congress from Utah from 2013 to 2023. Mark Beall is president of the AI Policy Network.