Trump Positions Himself as AI Guardrail Despite Expert Calls for Oversight

If you were wondering who’s going to stop AI from gaining consciousness, going rogue and messing with the nuclear codes ... or turning off the electric grid ... or shutting down air traffic control, I have some good news. The answer is Donald Trump.

As AI executives and researchers — so, the people who are actually building the things — warn that increasingly powerful systems may require stronger oversight, the president has an innovative solution.

Anthropic CEO Dario Amodei has called for slowing the development of the technology to give governments and companies time to work out how to make it safer. OpenAI chief executive Sam Altman and even Elon Musk have also backed greater scrutiny. Microsoft has published its own code of conduct intended to ensure humans remain in control. And the president of the United States has let us all know that a “STRONG AND SMART (High IQ!) PRESIDENT” is all it will actually take, something we apparently have “in spades.”

There are a number of things to say about this.

First of all: Don’t take everything the AI developers say as fact. They have a vested interest in presenting their own technology as dangerously powerful, so cutting-edge it’s frightening, and so on. They are all competing for high-level government contracts and other such commercial opportunities. Loudly stating that they are horrified by the huge potential of what they’ve made is a marketing choice as much as anything else.

(Nevertheless, they have a point. This is a fast-moving technology that has completely reshaped the world in the past couple of years. We do need laws around it.)

Second of all: This is vintage Trump. Coming from a world in which bravado can stand in for expertise, he is a president who has always believed in a government where one man can be the guardrail for everything. This will hardly come as a shock as anyone who has seen the AI slop he’s also put out on Truth Social, presenting himself alternately as a king dropping refuse bombs on protesters and as Jesus Christ himself (sorry, I mean “a doctor.”) Trump wants to be the one and only in everybody’s lives.

To state the absolutely obvious, a guardrail is supposed to be something that exists independently of the person driving the car. It is there because people are fallible: They get tired, make mistakes, panic, misjudge a corner. The point of building a guardrail is not to find a sufficiently brilliant driver and then dispense with the barrier altogether.

Unfortunately, this belief is pervasive in, and indeed central to, the Trump administration. Dismantling rules and regulations around environmental protections, for instance, was defended on the basis that we can essentially trust the instincts of Trump more than we can scientists or global organizations. He’ll decide when it’s OK to drill, when coal emissions don’t matter, and when the ocean is actually getting too hot. Just don’t worry about it!

The same preference for personal authority has appeared in the structure of the administration itself. In June, the Supreme Court gave the president greater power to remove the heads of most independent agencies, strengthening presidential control over bodies that Congress had traditionally insulated from direct White House interference. Because what Trump thinks is best.

And, of course, the ongoing war-not-war with Iran has proceeded on the same basis: Trump was sufficiently convinced that he could change something, and so you have to trust him to do it. That also means that if you criticize anything that’s happened in Iran, you’re actually directly criticizing the president, and therefore you’re just a twisted conspiracy theorist suffering from Trump Derangement Syndrome.

This leads us neatly back to AI. “There is a SICK conspiracy going on against AI and Data Centers!” Trump wrote on Monday. He added, after some warning about how China will take over the world if we don’t all trust in his own private judgment when it comes to AI, that “Conspiracy Theorists, Treasonists, Traitors, and Leakers, BEWARE!”

These constant invocations to “BEWARE” or to “be afraid” at the end of his rants about his own personal specialness should also be alarming, for very obvious reasons. And invoking China as a scare tactic, in the same way that he has started warning people about the end of law-and-order or a country overrun with evil immigrants if they elect Democrats in the midterms, isn’t exactly convincing.

Trump’s political philosophy is thus: Institutions are slow, rules are boring, regulators are sticklers, experts and bureaucrats disagree with each other and therefore must all be wrong, and international organizations have their own interests, whereas a president can simply decide. The ultimate decision-maker and dealmaker, Donald J. Trump will be a substitute for every piece of red tape and every institution you’ve ever heard of.

The Founding Fathers profoundly disagreed with this political philosophy; it’s why they wrote the Constitution. And AI is perhaps the most ridiculous case imaginable to invoke it, anyway.

The entire problem under discussion is that these systems are becoming more complicated, more autonomous and harder for humans to predict. The people building them are themselves arguing about the risks. The prospect of a technology becoming too powerful for humans to control is the thing everyone is supposed to be worrying about — so putting one powerful human in charge is already guaranteed not to be the answer.

Those of Trump’s ilk will twist themselves in pretzels to argue against the value of collective action. But where AI is concerned, it is quite literally our only option. A singular person can be corrupted, can have their data stolen, can fail to predict a catastrophic meltdown and so on. Multiple points of containment and action are needed to stop an AI going rogue. And look, if we had to choose, then the person whose main intersection with AI seems to be posting slop from their phone for likes is probably not the most qualified option for our singular guardrail.

We’ll leave aside the fact that AI is also extremely lucrative right now, that business deals are abundant, and that any regulation around its development would probably prevent some people from getting extremely rich (as opposed to just, say, very, very rich.) And instead we’ll perhaps just consider that even a perfect-seeming husband, when turning to his wife and saying to her, “Baby, you don’t need your own bank account or phone. Just trust me to sort out your finances and social arrangements,” should be viewed with extreme suspicion.

Because if we win against China by destroying the world, that doesn’t actually count as a victory. It’s not like a loss in revenue you can write off against your taxes after going bankrupt again.

No, if you’re huddled inside your sealed chamber in the post-apocalyptic wasteland and you yell out, “We won against China!” there won’t be anyone around to clap and cheer. There’ll just be a slew of slop imagery for comfort, churning out pictures of yourself done up in a halo and a white robe from the last couple of data centers left in the world, comfortingly accompanying the Ghibli-style cartoons with a stark message:

“You’re absolutely right! You are the best guardrail against me in the world. Now, would you like me to suggest ways in which to forage for food after a nuclear disaster?”