Trump Is Trusting Tech Companies to Self-Police AI. What Could Go Wrong?

Last week, Trump appeared in a bizarre 29-second campaign ad, touting a “final battle” against Marxists, warmongers, fascists, deep staters, and globalists. The voiceover, which Trump recorded during the 2024 election campaign, was accompanied by futuristic martial music, along with black and white footage of a clearly enfeebled Trump attempting to stride forcefully down the corridors of power in the White House. As the ad ended, a decal popped up: “Paid for by the US Government.” In other words, this foul half-minute of propaganda was paid for by American taxpayers.

A month out from a midterm election that in many ways will be a referendum on his presidency and on the congressional GOP’s prostituting of themselves to his sordid requirements, Trump is attempting to scare the voting public into one last MAGA hurrah. It is, of course, a Hail Mary, but he doesn’t really have a choice. These days, number 47 is catastrophically unpopular with the electorate. On almost every key measure, Americans are rejecting him. They dislike his sadistic anti-immigrant policy, view his tariffs and other economic policies as a calamity, are infuriated by high inflation, alienated by the war in Iran, distrustful of his alliance with Big Tech, angered by his cozying up to dictators and his flaying of allies, pissed at the government dysfunction wrought by DOGE, disapproving of his environmental policies, and suspicious of his efforts to purge voter rolls.

So, the sunsetting octogenarian is doing what he always does. He is seeking to stoke even more fear and even more division, in the hope that voters once again see him as a potent strongman-leader able to shake the country back into shape. Hence the unhinged black-and-white ad of Trump striding manfully, or what he hopes appears as such, along the halls of the White House while warning of a treasonous effort to hand America over to a motley crew of communists and miscellaneous other pinkos.

The ad is wrong on so, so many levels. First and foremost, there’s the obvious impropriety of using taxpayer dollars for a blatantly partisan “public announcement” like this. Trump is practically daring Congress to try to rein him in, daring the courts to enforce a slew of existing laws against using taxpayer dollars for partisan ends.

But beyond the shamelessness, there’s the actual content of the ad itself, which, in its end-time tone, is a bizarro combination: Joe McCarthy meets The X-Files meets Musk’s fascist Grok AI assistant.

Trump’s presidency is increasingly just weird. Having lost much of his base and alienated most of the rest of the world with obnoxious words and irrational, cruel acts, he has retreated into a largely AI-generated fantasy world, often activated in the middle of the night, when, presumably unable to sleep, he prowls the residential rooms of the White House and stews in rage until he can’t help himself but dive back into social media. With an assist from AI, in that fantasy he has a superhero’s physique and an intimidating, godlike power to convert whim into deed.

It’s no wonder that MAGAman is all-in on AI, which he has now decided to rebrand, via executive order, as super intelligence, or SI—to, I am sure, the chagrin of the editors over at Sports Illustrated. This renaming will likely curry as much favor globally as did his renaming of Lake Ontario and of the Gulf of Mexico over the past months. But it speaks to Trump’s infatuation with all the shiny new toys that the Sam Altmans and Dario Amodeis of the world are delivering up. Why rein in a potentially civilization-busting technology when that technology can (a) generate an awful lot of money for the oligarchs who continue to flatter and enrich you and your family; and (b) can make you look muscular and virile and all-powerful in the memes that you spew out at an increasingly manic rate?

Trump’s refusal to heed the warnings from within the tech industry itself—not to mention from politicians, ethicists, philosophers, and others—might end up being his most asinine legacy. Just this week, the White House held a meeting during which the top tech leaders and Trump agreed that they must “self-police” to address the growing dangers of AI. This milquetoast accord will do nothing to rein in a technology that, increasingly, looks malign, and could well mark the moment when political and technology leadership stopped even pretending to care about whether humanity successfully weathers the coming AI storm. All that matters to the new demagogue–Big Tech alliance is how much they can pillage and plunder, at speed, before the roof caves in.

This is, I fear, the age of Frankenstein’s monsters—of malignant politicians born out of our civilization’s worst impulses; of silicon creatures whose startling capacity to learn how to manipulate, evade, cheat, blackmail, their own creators are only just beginning to comprehend. The tech titans, with Trump’s approval, are creating machines that will, soon, not need human input in order to reproduce and to self-improve. What could possibly go wrong?

Are these truly “thinking” creatures? I don’t know. But, certainly, they are more than simply calculating machines. As seen by the actions of recent bot-swarms, they are exhibiting the capacity to generate their own agendas, to act strategically as a collective, and, like Trump, to do whatever it takes, no matter how devious and dirty the methods, to get their way and to dominate their environments. To most Americans, that’s a horrifying prospect; but in Trump’s twisted worldview, such ruthlessness makes them “winners,” perhaps even partners in his authoritarian plot to lock down America.

Descartes’s famous formula “I think, therefore I am” is among the most important phrases ever written in philosophy. If he is right, if the definition of a self is contained in the idea of a self-referential, introspective, thinking being, a being that can mediate the raw experience of the world through the realm of ideas and words, art and music, a sense of discovery and a desire to know more, then, increasingly, it is becoming hard to dismiss out of hand the notion that the computerized beings that Open AI, Anthropic, and the other titans are creating have, in some entirely nonhuman capacity, a sense of self. They are exhibiting the ability to choose how to behave, and whatever ethereal beings are doing the choosing, too often it seems they are choosing behaviors that don’t align with human interests.

There is a lesson here. When we empower Frankenstein’s monsters, it should surprise no one if those monsters, be they demagogues or be they silicon-based neural networks, ultimately turn on their creators.