Trump's 'unfettered AI development' is a race to the bottom

By Josie Calloway · Reporting from Pittsburgh ·

# The Quietest Patient

I spent a decade in ERs in Charleston and Pittsburgh. Triage follows one rule: the quietest patient is usually the sickest. Now, the loudest voices in the room claim everything is fine while the monitors flatline.

President Donald Trump is placing a deadly wager with the human species from a golf course in Ireland. According to reporting from CNA, Trump dismissed experts who warn about artificial intelligence as "very negative forces" who imagine scenarios that "won't happen." He wants to build AI without brakes. His logic is a hammer: "Whoever wins AI wins," and he will not let China lead.

The engineers building these machines are terrified. The BBC reported that Dario Amodei, the CEO of Anthropic, wants a global speed limit on the technology. Amodei said AI could erase one out of every two entry-level roles for accountants and paralegals, a shift that will "test who we are as a species." His team lead, Evan Hubinger, wrote on X that he sees a one-in-ten chance that AI kills every human within a decade. Jacob Coxon, a former researcher, resigned because he believes the builders are gambling with our lives.

When the people who hold the keys to the lab say the building is on fire, you don't call them negative. You leave.

The Dead Hand is Still Pulling the Trigger

The pattern is old, and it ends in a body count. This strategy mirrors the Cold War arms race, specifically the logic of the "Dead Hand." That mindset holds that maintaining a technical edge over a rival is the only way to stay safe, which turns every safety precaution into a strategic weakness. During the Cold War, that logic produced automated systems designed to launch nuclear missiles even if the leadership was already dead. The "win" meant destroying the enemy faster than they could destroy you.

Trump applies that same logic to AI. He views any speed limit as a surrender to China. He used this same framing during the trade war, where he used the goal of economic dominance to block restrictive regulations. It is a sprint toward a cliff, but the runners are too obsessed with their pace to look down.

The administration used a similar gamble during the invasion of Iraq. Officials dismissed expert warnings and argued that waiting posed a greater threat than a catastrophic mistake. We know how that triage ended.

Why the Security Shield Fails

Silicon Valley hawks and the administration argue that if the US slows down, an authoritarian regime like China will seize the lead. They claim a world where Beijing controls the most powerful algorithms is worse than a world with risky American AI. This argument works only if the goal is the power to shut down a rival's power grid.

The argument collapses if the scientists are right. If AI can truly hack any encrypted system or help a rogue actor design lab-grown pathogens that target specific genetic markers—risks the Pentagon and Anthropic researchers have flagged—then "winning" the race is a hollow victory. You cannot be the dominant superpower of a dead planet.

The friction between competing for profit and surviving as a species is not new. It is the same tension that nearly gutted the Paris Agreement. In both cases, leaders treated the atmosphere or the existence of the species as less important than the quarterly profit margin.

Now, the administration attacks the messengers. Trump designated Anthropic as a "supply chain risk" and a "radical left, woke company" because the firm refused to let the military use its models. When a leader cannot refute the data, he attacks the person presenting the chart.

A Sprint Toward the Edge

A collision is coming. Senator Bernie Sanders is co-sponsoring the "Ban Artificial Superintelligence Act" to force a pause. Meanwhile, a crop of CEOs—including Nvidia’s Jensen Huang, who said extinction risks are "complete nonsense"—are accelerating.

History suggests the unfettered approach will win because it pays more in the short term. At the 2023 AI Safety Summit, the call for a speed limit vanished beneath the drive for faster chips and stealthier code.

If we treat the risk of extinction as a "negative force" to be ignored, we aren't leading the world; we are just the first ones to jump. The Trump administration will likely continue to squeeze Anthropic and strip away every guardrail to ensure the US "wins," regardless of who dies in the process.

The victory Trump chases is a ghost. No one wins a prize for being the most sophisticated country in a graveyard.

Sources

  1. CNA: Trump says 'very negative forces' raising exaggerated concerns over AI
  2. BBC: Dramatic insider warnings over AI fall flat with some in Silicon Valley
  3. BBC: Dramatic insider warnings over AI fall flat with some in Silicon Valley