The Bat Signal

By Nick Catoggio

Tuesday, September 15, 2026

 

The New York Times published a poll this morning that has Democrats ahead by 8 points on the generic congressional ballot and by varying margins on seven of eight major issues tested.

 

Without looking: Which issue do Republicans still lead on?

 

Healthcare? C’mon. The GOP hasn’t had a thing to say about that in 15 years apart from “repeal Obamacare” and “cut Medicaid.” Democrats lead by 24 points in the Times survey.

 

The deficit? That’s a better guess, as Republicans have postured for decades as the party of fiscal responsibility. But their unearned and undeserved reputation on that point has at last fallen to ashes, incinerated by Donald Trump’s reckless spending. Democrats lead by 10 points.

 

Oh, I know—immigration. This would have been my guess, as I’d have assumed that four years of porous borders under Joe Biden would have put his party in a hole it couldn’t soon escape from. But no. In the span of 20 months, ICE’s goon-squad tactics and the White House’s gratuitous cruelty have erased the Republican advantage. Democrats now lead by 7 points.

 

That’s the whole poll in a nutshell. From immigration to affordability to foreign policy, the issues that propelled Trump’s comeback in 2024 have turned toxic for the GOP. Biden-era inflation hasn’t prevented the Democrats from building an 8-point lead on the cost of living amid an oil crisis and tariff-palooza. And Biden’s support for Israel’s unpopular campaign in Gaza is now a distant memory after Trump’s Iran misadventure, with Democrats ahead by 7 points on the issue of “foreign wars” and by 11 on the Israeli-Palestinian conflict.

 

Somehow, across nearly every national policy priority, the most ostentatiously populist president in modern American history has placed himself on the wrong side of The People. It’s genuinely amazing. Someone should write something about it.

 

It’s time to solve the mystery, though. The single issue on which Republicans still lead Democrats is—ta-da—artificial intelligence.

 

It’s a modest lead of 2 points, but a lead’s a lead. Why voters trust the right more than the left on the subject is, I suspect, mostly a function of Trump’s friendly relationship with Silicon Valley and with Elon Musk in particular. The average Joe’s thinking probably isn’t much more complicated than Elon knows what he’s doing, and Elon has influence with Trump, so Republicans will know what to do about AI more so than Democrats will.

 

The GOP comes to this subject with a built-in advantage, in other words, just as it did with immigration and affordability and foreign policy in 2025. The only question is how long it’ll take our tone-deaf president to demonstrate his genuinely amazing talent for squandering it.

 

The answer: not long!

 

Hoax.

 

Numerous polls published this year confirm that Americans are bearish on AI.

 

An NBC survey taken last month found a 31-69 split when respondents were asked if they’d support or oppose a data center being built locally. A Pew poll conducted in February saw 63 percent say AI is advancing too quickly compared with 21 percent who believe it’s advancing too slowly or at the right pace.

 

And when Quinnipiac asked people in March whether they think AI will do more good than harm in their day-to-day lives, opinion ran 34-55—down from 38-44 a year before. No less than 80 percent in the same survey said they were at least somewhat concerned about the technology.

 

To be clear: That’s what public opinion looked like before last week’s “we’re all gonna die” panic driven by employees at Anthropic, who alleged that AI could soon threaten humanity if developers don’t proceed more cautiously. The uproar led Anthropic CEO Dario Amodei to publish an essay Saturday calling on competitors to join him in “pacing the frontier,” i.e., slowing down, introducing third-party evaluators into their operations to police risk-taking, and cooperating with each other to set safety standards.

 

OpenAI’s Sam Altman quickly chimed in, saying that he agreed. So did God-emperor Elon. Between the public’s baseline skepticism about AI and the tech-lord consensus that progress is moving irresponsibly fast, there’s no doubt that “pacing the frontier” will be the position of most Americans on the subject—and that they’ll expect a degree of federal oversight of the process. “Some forms of coordination that would be impactful for pacing are legally challenging, and will require government support,” Amodei warned in his essay. Regulation of some form is inevitable.

 

Populists as far afield as Sen. Bernie Sanders and Steve Bannon have aligned themselves with that position. So have Democrats who know a thing or two about winning national elections. But Trump? With a majority rapidly forming in favor of caution on AI, the president entered the fray yesterday and planted his flag (where else?) firmly on the opposite side of it.

 

Across at least seven(!) different Truth Social posts, he attacked the idea of slowing the technology’s development. “The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the U.S.A. has that, in spades!” he wrote in one, not at all reassuringly. He compared anxiety over artificial intelligence to anxiety over global warming, a clever pander to right-wingers who are still making up their minds on the issue but a poor analogy. In the climate change debate, it wasn’t the fossil-fuel industry itself that led the effort to convince the public that its product is dangerous, was it?

 

Again and again in his posts, the president used the same word to describe the alleged risk of a renegade AI: hoax. Which caught my attention, as that term isn’t used lightly in the parlance of Trump-ese. It has a particular meaning and is designed to produce a particular result.

 

“To MAGA true believers, the word seems to function like a rhetorical Bat Signal,” I wrote last year in a different context. “When a ‘hoax’ is afoot, populists are expected to rally behind their president, no questions asked…. [It’s] essentially an incantation by which Trump’s political operation turns very bad political developments into loyalty tests for their ultra-loyal base.”

 

Hoax! is what the president cries when he feels cornered on a subject and fears he might lose a meaningful portion of his party over it. It’s the closest he gets to speaking ex cathedra to his cult: When he uses the H-word, the matter in question instantly becomes a matter of Us vs. Them in which believers are expected to adopt his position as dogma. His two impeachments, his election defeat in 2020, even the contents of the Epstein files: If you’re credulous about any of it, you’ve fallen for a hoax and betrayed the faith.

 

He’ll brook no dissent on such matters. In the case of Epstein, he went as far as to excommunicate supporters who persisted in wanting to see what was in the Justice Department’s archives after Trump flashed the Bat Signal at them to ignore it.

 

The fact that he immediately leaped to categorize caution about AI as a “hoax” tells me that he knows he’s in the minority here, is nervous about it, and is desperate to force rank-and-file Republicans into mindlessly backing his “full speed ahead” position before they have a chance to form their own opinion on the merits. But that brings us to a new mystery: It’s weird that Donald Trump, of all people, should be so reflexively opposed to regulation, no?

 

It’s not weird in the sense that he’s some Milton Friedman disciple who’s normally doctrinaire about the magic of free markets. Just the opposite. I’d expect Trump to salivate at the thought of his administration being invited to meddle in the most dynamic and consequential industry on Earth. This is a guy whose government has taken ownership stakes in dozens of private companies and who just mugged Venezuela into giving the United States a huge share of its oil. Now Anthropic and OpenAI are demanding that the feds get involved in their operations, and his response is “How dare you?”

 

Isn’t that weird?

 

It’s the economy, stupid.

 

It’s not so weird under the circumstances.

 

It all starts with the economy. Despite the public’s discontent with 20 months of Trumponomics (Democrats lead by 7 points on the economy in the new Times poll), I suspect the president’s information bubble is thick enough that he believes he still wears the economic halo he earned during the boom years of his first term. The latest jobs report was terrific, wasn’t it? We haven’t had a recession, have we?

 

More so than with any other issue, Trump seems convinced that his presidential legacy will rest on how the economy performs during his tenure. He appears willing to try almost anything, no matter how foolish—striking a terrible peace deal with Iran this summer, pressuring the Federal Reserve to slash rates while inflation is running hot—to reduce the risk of a slowdown. His greatest fear is being remembered as the Herbert Hoover of his era.

 

When your politics boils down to “growth at all costs” and you also happen to be a born sociopath, the debate over artificial intelligence is a no-brainer. The stock market and increasingly the wider economy depend on sustaining the AI boom; that’s likely all Trump needs to know to support letting development continue unchecked. Who cares if there’s a 10 percent chance that the technology will kill everyone eventually—particularly if it’s unlikely to happen before he leaves office in 2029?

 

He’s happy to increase the risk of a mass extinction during J.D. Vance’s presidency to avert a recession during his own.

 

Growth, growth, growth explains 90 percent of Trump’s “full speed ahead” approach to AI, I’d estimate, but Axios rattled off a few other theories this morning, some of which are more plausible than others. “A historic buildout of data centers, chip fabs and power plants—physical monuments to investment and growth” surely does appeal to a guy who devotes most of his mental energy to buildings. Ditto for the president bristling at “an elite ‘expert’ class telling him no” on AI acceleration. It’s distinctly possible that the man in charge of America’s national security wants to throw caution to the wind on an existential threat for no better reason than that Barack Obama thinks he shouldn’t.

 

But other explanations are harder to swallow, such as Trump supposedly caring about maintaining the lead in “a race with China that America is currently winning,” as Axios speculates. Since when does the president regard Beijing as an enemy to be defeated, especially on this issue? If he cared about outpacing the Chinese on AI, he wouldn’t have allowed them to buy advanced American chips, right?

 

By wrecking U.S. alliances and governing like a banana-republic caudillo, Trump has turned his entire second term into an exercise in advancing the cause of Chinese global hegemony. Why would he undercut that cause with respect to artificial intelligence?

 

Pro-oligarch.

 

If there’s any core reason apart from economic growth that explains the president’s hostility to regulating AI, it was his pal Bannon who inadvertently touched on it. “American citizens cannot be and will not be supplicants to a handful of tech oligarchs,” Bannon proclaimed to Politico in making the case for federal regulation. That’s a fine populist position but an awfully tricky one for him to take in 2026. Because, you see, Donald Trump loves oligarchs.

 

All authoritarians do. Oligarchy suits their instinctive third-world preference for patronage, unaccountability, and influence peddling. Having loyalists and cronies in charge of fantastically lucrative private industries makes it simple for the caudillo to influence those industries and to wet the beaks of various players across society as needed to get his way. Power is easier to wield when it’s concentrated.

 

Plus, there are lots of fun perks in having rich friends with financial reasons to keep you happy.

 

America’s tech lords rushed to signal their allegiance after Trump was reelected, realizing that their fortunes would now depend in part on maintaining the leader’s favor—another hallmark of oligarchy. Some mega-bucks Silicon Valley mainstays, most notably Musk and David Sacks, joined the president’s reelection campaign and were rewarded with administration positions after he took office. Sacks, the co-chair of the President’s Council of Advisers on Science and Technology, has reportedly leveraged that influence to shape Trump’s hostility to “pacing the frontier” of AI development: The president’s views on the matter are “all coming from him,” one White House official told Politico.

 

To Trump, it must have been mortifying watching Amodei and Altman reject an oligarchical relationship with his administration in favor of traditional regulation by Congress. That’s what explains his seemingly surprising “how dare you?” attitude toward “pacing the frontier,” I think. Since when do private stakeholders appeal to the legislature for help in what’s supposed to be an autocracy?

 

The whole point of postliberalism is to dispense with formal rules that constrain how the government and its subjects behave and instead let the powers that be reward “friends” and punish “enemies” unaccountably. Rather than be “friendly” and deal with the White House informally, the AI leaders opted to demand accountability by getting the rest of the government involved. It’s tantamount to a Mafia don dangling an offer to partner with someone in a new racket and having that person call the cops in response.

 

In an oligarchy, the oligarchs’ power is the leader’s power. “Pacing the frontier” is a rare case of oligarchs offering to forfeit a degree of power for some greater good. Of course the leader is unhappy with them.

 

The strongest case for Trump’s position was made by an oligarch’s protege, aka Peter Thiel’s boy, aka the vice president of the United States, who told reporters that pleas for government intervention by Amodei et al. looked suspiciously to him like a “Trojan horse.” J.D. Vance was alluding to the possibility of “regulatory capture,” in which a powerful firm tries to cement its competitive advantage by encouraging the government to impose burdensome rules on its industry. A company that’s already well established can bear the exorbitant costs of complying with those rules and knows that fledgling rivals who might one day surpass it cannot.

 

That could be what Anthropic and OpenAI are aiming for here, creating a regulatory barrier to entry for startups that might otherwise threaten their dominance over AI and concocting a cynical “we’re all gonna die” public panic to bolster the cause. If they want to “pace the frontier,” one might ask, why do they need the feds to step in? Nothing’s stopping them from agreeing among themselves to do so, right?

 

I don’t think regulatory capture is what they’re after, though. For one thing, there sure do seem to be a lot of AI engineers who are worried about the threat. (Chinese intelligence is worried too—although, true to form, officials are worried about the safety of the Communist Party, not of humanity.) And I don’t see how firms agreeing to police themselves for safety risks, without being compelled by law to do so, would solve the collective-action problem the industry faces. Absent legal compulsion, a startup desperate to catch up to Anthropic logically won’t police itself in its haste to close the gap with its rival.

 

In any event, “regulatory capture” isn’t going to win the battle for public opinion for the White House. Neither is the “hoax” Bat Signal. Soon Republicans will trail Democrats on every issue that American voters deem important, extinguishing Trump’s last bit of populist credibility. Never mind the economy: That’s his legacy.