Whatever beef Donald Trump had with Anthropic CEO Dario Amodei seems to have been buried. The two had dinner on Sunday, followed by a luncheon with other CEOs who then stood together with top Republicans outside the White House. Together, they announced a “morally binding” agreement about the governance of artificial intelligence, or “super intelligence,” as Trump has rebranded those companies’ products. While Amodei, Sam Altman, and Elon Musk had purportedly butted heads with the president in recent weeks over the idea of slowing down the launch of new and more capable large language models—the executives called to “pace the frontier”; Trump disagreed—they have reached a new understanding. “We all need to work together to make sure that we can win, and we can win safely,” Amodei said, before slinking awkwardly back into the high net-worth clump of suits assembled behind him. “If we can do this right, work with the President and everyone here, we can win safely.” Trump summed up the principles of their 308-word agreement more succinctly: “tremendous self-regulation.”
Since Trump took office, pundits have marveled at the tech industry’s rightward shift. The recent explosion of discourse over “AI Safety” has reenergized that conversation, as observers question how lib-coded ostensible true believers in regulation such as Amodei—who publicly fought with the White House over the terms of its defense contracts—could make peace so easily with the avatar of the reactionary right. Perhaps his wincing at Tuesday’s press conference betrayed some inner torment. But Amodei isn’t the only self-styled do-gooder to make nice with Trump. Just a few years ago, Mark Zuckerberg—who could barely contain his glee while standing next to the president—poured well-publicized millions into liberal causes. Jeff Bezos and Elon Musk cultivated their image as climate champions. It wasn’t so long ago that Sam Altman called Trump “unfit” to be president. “i guess the meme is that Dario is awkward, but it is kind of weird how eager the other CEOs are to participate in a public knee-bending ceremony like this,” tech journalist Kevin Roose marveled of Tuesday’s White House summit with Silicon Valley. “you’d hope more of them would be visibly uncomfortable!”
What explains the new coziness? The terms of the White House’s truce with tech leaders aren’t exactly mysterious. The same executives have showered Trump’s SuperPAC and pet projects with $227 million, the watchdog group Public Citizen notes. There are obvious benefits to having the administration’s enthusiastic support. But there’s more than just opportunism at work here. For as much has been made of the cultural and ideological fissures between MAGA and Palo Alto, the two camps share a common quasi-religious faith that economic growth can solve virtually every problem known to man.
And although there are certainly weirder corners of the overlapping ideologies that buzz around Bay Area offices and polycules, even top “true believers” who’ve trafficked in rationalism and effective altruism have been steadfast about their belief in growth as a somewhat mysterious but ultimately positive force that promises untold fortunes to those who choose to feed it. Way back in 2013, Sam Altman wrote an essay making the case that rapid economic growth “is not only critical for startups, but for most systems. Either you’re growing, or you’re slowly dying.”
In the essay, Altman argued that attempts to mitigate the inequalities created by the twenty-first century’s biggest growth industries (“robots”) could easily backfire. “We should understand,” he continued, “that as a consequence of technology and an economy of ideas, the gap between the rich and the poor will likely increase from its already high-seeming levels. There is good and bad to this, but we should be careful not to legislate against it, which will hurt growth.”
This kind of cold supposed truth-telling isn’t exactly novel among founders. Scottish philosopher and effective altruism guru Will MacAskill famously counseled disgraced crypto executive Sam Bankman-Fried that the most effective way he could be altruistic was to make as much money as possible. The idea of “earning to give” is obviously compelling for people who would like to become wealthy, and those who see an opportunity to offer those people lucrative moral guidance. The idea has been something of a mainstay for effective altruists, who like to make up numbers to justify why fortunes that also flow to their pet causes, including “AI Safety,” are sum-positive for the cause of building a more abundant world.
To find the origins of this kind of reasoning you have to go way back. In 1889—at the height of the Gilded Age—industrialist Andrew Carnegie made the same case. “Not evil, but good, has come to the race from the accumulation of wealth by those who have the ability and energy that produce it,” Carnegie wrote in The Gospel of Wealth. That rich men had amassed such incredible fortunes testified to their “superior wisdom, experience, and ability to administer” those fortunes to worthy causes. That spirit of far-sighted charity would be a cudgel against “Socialists and Anarchists” looking to upset what Carnegie saw as the natural order of things—the socialists and anarchists who also happened to be organizing at his mills. “The laws of accumulation will be left free; the laws of distribution free,” he wrote. “Individualism will continue, but the millionaire will be but a trustee for the poor; entrusted for a season with a great part of the increased wealth of the community, but administering it for the community far better than it could or would have done for itself.”
Preachers of today’s newfangled Gospel of Wealth ascribe utopian potential to the stuff that’s making them rich, too. Pressed to articulate the benefits of artificial intelligence—and not just its catastrophic risks—Amodei’s core argument has been that large language models will deliver not only technological breakthroughs but riches fantastical enough to save the world and solve its most intractable problems. In his long blog post last month about pacing the frontier, Amodei argued that his industry’s products “could cure most major diseases in the next 5–10 years, greatly accelerate economic growth rates, create a world of abundance and empowerment, and usher in a renaissance of democracy and freedom.” His “dream scenario—perhaps a goal to aim for—would be 20% annual GDP growth rate in the developing world, with 10% each coming from AI-enabled economic decisions and the natural spread of AI-accelerated technologies.”
On Tuesday, Trump echoed Amodei’s excitement about LLMs’ capacity to deliver “unbelievable growth.” Faced with rising inflation, mortgage rates, and sprawling blowback from an unpopular, illegal, and expensive war on Iran, Trump and Treasury Scott Bessent have both been adamant that the United States can simply grow its way through economic headwinds. That faith rests in large part on the men who gathered outside of the White House on Tuesday. Investments tied to the building binge on chips and data centers have accounted for about 20 percent of the United States’ economic growth this year. Frothy valuations continue to buoy the stock market and, with it, the fortunes of the wealthiest 10 percent of Americans, who are responsible nearly half of all consumer spending. As eager as Silicon Valley executives are to stay on Trump’s good side, Trump wants to keep them happy, too—and to appease voters angry about the risks of data centers and artificial intelligence alike.
Trump and Silicon Valley executives, however, seem to share another core belief that helps inform this week’s “self-policing” spectacle: impunity. Given the steady drumbeat of headlines about OpenAI agents hacking into companies, nonprofits, and government agencies, it isn’t hard to imagine why self-policing is preferable to the alternatives. Altman’s 2013 ode to the powers of growth ends with some suggestions for how to increase growth, including reforming tort laws that can be used to try companies for reckless and negligent behavior. Altman writes that “legal protection is of course important, but it’s gotten so silly that it discourages innovation.” As former Federal Trade Commission chair Lina Khan wrote this week, an unenforceable pledge is certainly preferable to being held accountable under the many laws that already exist to punish corporate wrongdoing. There’s a chance that could happen; Khan’s successor this week announced that it was investigating Anthropic, OpenAI, and other unnamed companies over the potential dangers posed by their products.
Another, potentially more likely outcome is that Trump loyalists at the FTC clear those companies of any wrongdoing. Trump, for his part, seems to believe that holding companies to existing laws and regulations—much less creating any new laws and regulations—would threaten the AI boom. The people getting rich off it are inclined to agree.
