AI Has Broken Containment - The Atlantic

AI has ascended to the role of main character. When Donald Trump traveled to Beijing for an historic summit last week, AI was one of the central topics of his discussions with Xi Jinping. As the two nations remain locked in a technological arms race, the president brought along some of the United States’ most powerful AI executives, including Elon Musk and Nvidia’s Jensen Huang. A continent away, the European Union has been unsuccessfully petitioning Anthropic to grant access to its advanced cybersecurity model, Mythos. Back in the United States, millions of students and teachers are dealing with the fallout of a devastating ransomware attack on the software platform Canvas—a hack that was likely aided by AI tools. And on Thursday, Cisco became the latest major company to justify layoffs by pointing to AI.<br>The past six months have marked a sea change in the reach and influence of AI. For most of 2024 and 2025, there was talk of AI progress slowing down or even stopping altogether. Even as the technology began to infiltrate schools and reshape financial markets, AI was relatively easy to compartmentalize from other major, more pressing issues in American life.<br>No longer. Now the technology has become regarded as a matter of the greatest economic, political, and global consequence. The most important issues in U.S.-China relations? Tariffs, Taiwan, and AI, apparently. Political leaders and pundits including Bernie Sanders and Steve Bannon have put AI center stage, and the backlash against data centers is loud and inescapable. The specter of AI-driven layoffs hangs heavy—as does the threat of advanced hacking bots capable of taking down electrical grids and breaking into banks. All manner of once-speculative concerns about AI have become pressing matters. There is no longer a distant AI future so much as the mess we are all forced to confront today.<br>The newly chaotic and inescapable state of AI is the result of two inflection points. The first came at the start of the year, when AI agents exploded in popularity. Products such as Anthropic’s Claude Code and OpenAI’s Codex don’t just talk to you; they can do things on your behalf—code, trade stocks, analyze spreadsheets, generate slide decks, and even create Amazon listings. The technology’s once-questionable economic value became very clear, very quickly, to a large number of businesses, which have clamored to incorporate agents alongside, or in lieu of, their human employees. As agents have swarmed the workplace, nearly three-quarters of employed Americans think AI will decrease overall job opportunities and 30 percent of Americans are concerned that AI will make their own job obsolete.<br>The second shift began in late February. First, a high-profile contract dispute between Anthropic and the Pentagon revealed how essential AI has become to national security. Then, in early April, Anthropic announced Mythos, a model with the ability to rapidly find and exploit bugs throughout the internet. (Shortly after, OpenAI came out with an analogous model.) In tandem, these events suggest that some of the most catastrophic fears about AI could come true: Several independent cybersecurity experts have told me that these models are approaching the abilities of the most elite human hackers. Anthropic and OpenAI have not released these cybersecurity models to the public, out of fear they will be used by criminals or terrorists; meanwhile, companies and government bodies alike are hungering for access so they can use the tools to patch any bugs. As a result, AI labs have become major geopolitical actors in their own right.<br>Spurred by the threat of massive AI cyberattacks, the Trump administration is now reportedly weighing the possibility of testing or even licensing the most powerful AI models before their public release—moves the White House once called “dangerous” and “onerous.” White House Chief of Staff Susie Wiles is said to be spearheading Trump’s AI policy and has written a rare post on X vowing to keep Americans safe from AI cyberattacks by ensuring “the best and safest tech is deployed rapidly to defeat any and all threats.” (A White House official told me that “any policy announcement will come directly from the President.”) This month alone, dozens of members of Congress have signed letters to the White House on AI regulation.<br>It’s hard to overstate the extent to which AI has crept into contemporary life, even for people who aren’t commonly using the technology. A poll this spring showed that, for Americans, AI is growing in importance faster than any other issue. AI wasn’t a focus for campaigns in 2024, but several races coming up this year are poised to involve heated debates over the technology. Data centers in particular have gone from basically invisible to a divisive issue that cuts across party lines: 70 percent of Americans oppose the construction of an AI data center in their community. These centers’ voracious...