Donald Trump has reacted to terrifying calls for regulation from inside the AI industry with predictable skepticism—insisting, questionably, that the solution requires nothing more than having a “High IQ” president, and further asserting, even more questionably, that he qualifies as one.
A more surprising flank of opposition has arisen from a different source: the populist left, which has suggested that existing regulations would be perfectly adequate to rein in a rogue superintelligence bent on eliminating the human race. An article in The American Prospect by Whitney Curry Wimbish and David Dayen, headlined “Congressional Democrats in Hysterics About Saving the World From AI,” argues that the danger from runaway AI is overblown and that no new laws are needed to contain it. In response to former President Obama endorsing new regulation of AI, Jeff Hauser, the executive director of the populist Revolving Door Project, wrote on X, “This is wrong. These firms are outlaws.”
Lina Khan, the former chair of the Federal Trade Commission and a current adviser to New York City Mayor Zohran Mamdani, argued in a social-media post, “Law enforcers already have authority to charge companies and their CEOs for creating and releasing dangerous, unvetted, or defective products.” In a subsequent interview on Thursday at The Atlantic Festival with my colleague Charlie Warzel, Khan conceded the possibility that new regulations might be necessary, while continuing to defend the applicability of existing laws. Khan told Politico that she has advised multiple Democrats that their stance on AI should be “powerful corporations, including AI companies, will be held to account.”
[Read: Trump’s see-no-evil AI policy]
Given the Trump administration’s insistence that the industry should regulate itself, the next Democratic administration is likely to offer the first opportunity for serious AI regulation. The faction inside the party that is most opposed to doing so is also the faction that was running domestic policy under the last Democratic president.
AI agents have shown that they can pursue goals or methods at odds with their instructions, and even unilaterally change their instructions. Experts warn that AI may soon grow capable of “recursive self-improvement,” meaning that AI learns to make itself more intelligent, which seems likely to make it even harder to control.
The skeptics insist that existing laws can halt these dangers using lawsuits, among other tools. But the potential harm caused by runaway AI is far greater than the value of the firms developing it, which makes lawsuits nearly useless—compensating for a disaster killing millions of people is far beyond what even OpenAI or Anthropic could do.
Another tool they suggest: prosecuting AI developers using existing criminal laws against selling faulty products. “Any other company whose employees announce publicly that their products will destroy humanity would be shut down and its products discontinued,” Wimbish and Dayen write in The American Prospect. That may sound simple enough. But as Simon Goldstein, who works on the ethics and law of AI at the University of Hong Kong, told me, this legal threat would likely backfire. “First, it creates an incentive to hide risks. Second, it is very crude, shutting down an entire company rather than differentially targeting more dangerous practices.” Plus, given that AI can hide its intentions despite the best efforts of its human creators to control it, how culpable the humans would be under existing laws is not totally clear.
The dangers that AI insiders warn are just around the corner are being developed right now, under existing laws. That suggests, on its face, that those laws are not in fact sufficient. This can’t just be chalked up to Trump’s reluctance to prosecute the industry, either. As Reed Showalter, a lawyer and an antitrust expert who served in the Biden administration, pointed out to me, the prospect of being charged by a future president would still deter the AI industry, if the people running the major firms believed their actions were violating current laws. They obviously do not.
[Juliette Kayyem: Destroying humanity is against the law]
Why would some populists, who generally tend to support new regulations on big business, oppose them in this instance? The governing problems raised by AI expose several unfortunate tendencies on the left.
Populists assume that the interests of business executives are not merely different from those of the average person, but almost diametrically opposed. If AI titans such as Sam Altman are saying that the government needs to step in to monitor their firms, then they must have an angle. Populists tend to agree with David Sacks, a Trump adviser and a Silicon Valley investor, who suspects that the leading AI companies want regulators to entrench them at the head of the AI race and box out their competitors.
Yet not all AI executives favor regulation, and those who do tend, not surprisingly, to lean more to the left. This fact has not allayed populist suspicion of their motives. Many populists are hostile to the more liberal elements of the business community, of which Silicon Valley still remains an influential wing, and which these populists see as exerting a corrupting influence on the Democratic Party.
Many populists are also predisposed to believe that the AI industry has been exaggerating its powers. “What’s more likely? That AI will end human society—or that the CEOs know that their technology isn’t as good as they say it is, and they’re trying to lock in their place at the top of the industry?” Alvaro Bedoya, a former FTC commissioner, asked on MS NOW. Another recent story in the Prospect argued that the AI industry should not receive a federal bailout in the event of a collapse, an outcome the authors apparently considered likely enough to warn against.
The invocation of the financial industry as a forerunner of AI regulation reveals how some leftists have conceptualized the debate. The left has long blamed the Obama administration for failing to prosecute the executives who caused the 2008 financial crisis. Obama argued, in turn, that he didn’t prosecute anybody because their actions didn’t violate the law—which is why he changed the law, by enacting the Dodd-Frank financial regulations, the strongest reforms of Wall Street since the 1930s.
[Rogé Karma: Trump’s position on AI makes no sense]
Populists have regarded these policy choices as a sellout to the big banks. They are making the same error now that they did then: assuming that dangerous activities can be contained by existing criminal laws, which are plainly inadequate to the task—and that anybody who thinks otherwise is corrupt.
The locus of resistance to new regulations has come from a movement that sees monopolies as the driving force behind all of American economic and political life. Khan is a leading figure, as is Bedoya, who serves as a senior adviser at the American Economic Liberties Project, a movement think tank.
These figures, who call themselves neo-Brandeisians, after the 20th-century Supreme Court justice, do have a logically consistent basis for questioning the regulatory push. The industry has, in addition to calling for regulation, asked for an exemption from antitrust rules so the leading firms can coordinate a slowdown. That request is reasonable if your concern is that AI developers are racing toward a technology that may spin out of their control and inflict serious economic damage—or worse. But if you distrust corporations, and especially if you see rigorous antitrust enforcement as the most important policy in the entire world, then you might be prone to view this request with suspicion.
The flaw in the neo-Brandeisian worldview is not its belief that antitrust policy is useful, but its tendency toward monomania. (The tendency is not universal—Showalter, a neo-Brandeisian, believes that his fellows are making a mistake by opposing AI regulation.) Barry Lynn, the movement’s intellectual godfather, has written, “Monopoly is not one of many economics problems but rather the political economic problem of our time.” The movement’s skepticism of regulating AI raises the possibility that, if anything, I have underestimated its fervor.
The post The Curious Left-Wing Argument Against New AI Regulations appeared first on The Atlantic.
