Join our FREE personalized newsletter for news, trends, and insights that matter to everyone in America

Newsletter
New

The Limits Of Positive Thinking

Card image cap

As anxious Americans dealt with a cascade of computer failures over the past week, Donald Trump offered this advice: “Relax.”

The air-traffic-control system, email providers, banks, airlines, and even the internal communications system of the country’s biggest intelligence agency had all seen outages, fueling fears of digital conspiracy. But they weren’t connected, the president assured Fox News. And hackers were not behind the outages, he insisted, because no one would dare risk retaliation by the United States. Companies were simply replacing outdated equipment, “and things happen when you put in new equipment,” he said. There was nothing to see here, no cause for alarm. “Everything’s good.”

Digital-security experts are not nearly so sanguine. They see something new and potentially ominous that’s fueling these hugely disruptive outages: artificial intelligence. Companies haven’t been replacing old equipment so much as rapidly patching newly discovered flaws in their systems with the help of sophisticated AI that can quickly spot vulnerabilities.

And why are companies moving so fast? Because attackers have AI too. The FBI warned this week about an AI-enabled tool set, called ARTEX, that “cybercriminals and state-aligned hackers” are using to identify weakness and develop attacks in under three hours. None of the affected companies has said that it thinks its systems went down because of an attack. But security experts told us that in the past few weeks, they have been struggling to keep up with an onslaught of attempted hacks from AI-enabled adversaries. And rushing to harden lots of computer infrastructure all at once can be messy and inconsistent, explaining some of the disruptions.

Trump has never fully reckoned with the risks of AI. He prefers to call it “superintelligence,” an odd choice given that that’s the term some doomsayers use for AI that becomes self-aware and tries to wipe out humanity. The president frames the race to develop advanced AI as an existential contest between China and the United States. “Whoever wins AI wins,” he is fond of saying. The Defense Department has deployed AI systems across its networks with seemingly little regard for the risks of using them in combat. Trump has avoided creating safety regulations and guardrails and has followed the counsel of executives whose fortunes depend on AI expansion. That includes ex–Nvidia CEO Jensen Huang, who has consistently downplayed the risk of out-of-control AI, calling apocalyptic scenarios “made-up” or baseless.

When it comes to developing and deploying the technology, Trump is all gas and no brakes.

For a different president, the nerve-racking outages of the past week might have been a moment for sober reflection on whether the United States is prepared for what’s coming. Even some Republicans are asking for that, as are some Democrats. “It feels like every day brings another reminder that AI is moving faster than our ability to manage its consequences,” Senator Mark Warner of Virginia, a Democrat who is in line to chair the Senate Intelligence Committee in the next Congress, said in a statement. Warner has proposed legislation that would subject AI models to greater scrutiny and safety checks. “Americans deserve clear, reliable information and a government that is prepared to respond,” Warner said.

But the administration seems not to grasp how worried many Americans are at suddenly being unable to check their bank balance or read their email. The disappearance of everyday conveniences comes after months of dark warnings about rogue AI agents and renewed fears of cyberattacks from Iran. The French government publicly accused Iran last week of remotely sabotaging food-refrigeration units and causing mass spoilage in grocery stores. Trump may insist that there’s nothing to be concerned about, but even some of his allies fear that something bigger is at play. “If this is all a coincidence,” Senator Josh Hawley of Missouri, a Republican, told reporters on Tuesday, “it’s like the greatest coincidence in the history of the world.”

Normally, in the middle of a domestic crisis, an administration tries to speak with one voice and tell a coherent story to an anxious public. But as bank websites crashed at the same time as stranded passengers slept in airports, no one in the administration seemed to know what was happening or who was in charge of fixing things.

On Monday, Homeland Security Secretary Markwayne Mullin, who is supposed to lead the government’s response to domestic cybersecurity incidents, suggested that artificial intelligence deployed on computer networks was “self-fixing itself.” In the scenario he described, recent network collapses were the result of a big, messy patching job. That is close to what experts think is continuing to happen. But Mullin attributed his understanding to a tech leader, whom he declined to name, telling Fox News that his theory “kind of makes sense.” This uncertain description, from an official with access to an array of classified government intelligence and seasoned briefers, did little to allay concerns.

[Read: The secretary of imprecision]

The unit within Mullin’s department that should be running point in this crisis, the Cybersecurity and Infrastructure Security Agency, has lost almost a third of its 3,200 employees since Trump returned to office two years ago. The White House wants to cut the budget for some agency programs, including election security, by $700 million next year. When we went to the CISA website today, we found guidance about recent incidents, forms for reporting cyberattacks, and a two-month-old handbook for “Communicating Under Pressure.”

“During service outages, clear communication matters,” CISA unironically advised.

Officials dismissed any connection between the computer failures and the deep staffing and budget cuts at the agency. “The implication that the recent events would not have occurred if CISA had more employees is patently false,” a White House official said.

The new director of national intelligence, Jay Clayton, told a congressional committee yesterday that “the timing of these ​incidents can raise questions, ​and it should,” but he insisted that the outages were coincidental and “associated with maintenance and ​technical ​changes.” Clayton also blamed the patching of vulnerabilities, but he didn’t blame AI.

Marco Rubio, the secretary of state and the national security adviser, appeared initially confused when a reporter asked at a press conference in Manila whether AI had played a role in the outages. He said he couldn’t confirm that that was the case but explained why AI hypothetically could have led to the systems failing. Rubio stressed that “people are fearful; people are uncertain” and said that it was important for the administration to “share with them transparently what we know.”

Trump’s habit, when he is confronted with a crisis, is to rely on the power of positive thinking. In the first weeks of 2020, he was slow to sound the alarm on a new virus discovered in China, reportedly for fear of rattling financial markets and upsetting Xi Jinping. By the time he left office a year later, some 400,000 people in the United States had died of COVID-19. A similar dynamic now seems to be at work. Trump has resisted calls even from within the AI industry to regulate the technology, fearing that government interference will stymie innovation and slow growth. The president knows that AI companies are propping up the stock market and that U.S. economic expansion depends largely on the construction of new data centers, which are essential for developing more powerful AI models.

Trump was in China this week for meetings with Xi and Vladimir Putin, and he indicated to allies that he didn’t want anything to distract from those sessions. That could explain the at-times-nonexistent messaging from the White House. During the height of Friday’s ground stop by the Federal Aviation Administration—which does not appear to have been caused by AI-related patching—Trump was huddled in the Oval Office with the actor Jon Voight, talking about Hollywood tax rebates. He stayed out of sight for days, leaving Mullin and Transportation Secretary Sean Duffy to make a widely panned series of appearances in which they largely blamed the Biden administration and demanded money from Congress. One air traffic controller told us that everything he and his colleagues had learned about the nationwide ground stop—an event that has happened only three times this century—came from the same general announcement Duffy had made to the public. The controller said that he had received no information internally about the outage and its causes.

Officials privately acknowledge that an extraordinary amount of work needs to be done to fend off the relentless and increasing AI attacks. But the president’s advisers may be wary of telling the public the truth about the scale of the threat because they don’t want to make Trump angry, two Republicans close to the White House told us on the condition of anonymity to discuss private conversations.

[Read: Trump sets the olive branch on fire]

In his Fox News interview, Trump said that journalists were exploiting the computer crashes for political purposes: “The press is sort of having a lot of fun with it. If these things would’ve happened two years ago, three years ago, in the Biden administration, which was grossly incompetent and corrupt, nobody would’ve mentioned it.” Journalists extensively covered similar breakdowns, including a nationwide ground stop in 2023 and the discovery of a massive penetration of U.S. telecom networks by China, through an actor known as Salt Typhoon, the following year.

Trump insisted on Fox News that the weeklong parade of cyber horribles was a mere coincidence. “I’ll let you know if there’s a problem,” he said.

After the past week, the bar for what Trump considers to be a problem is looking awfully high.

Isaac Stanley-Becker and Madison Hahamy contributed to this article.