One week ago, a new and formidable challenger for OpenAI’s throne emerged. A Chinese AI start-up, DeepSeek, launched a model that appeared to match the most powerful version of ChatGPT—but, at least according to its creator, was a fraction of the cost to build. The program, called DeepSeek-R1, has incited plenty of concern: Ultrapowerful Chinese AI models are exactly what many leaders of American AI companies feared when they, and more recently President Donald Trump, have sounded alarms about a technological race between the United States and the People’s Republic of China. This is a “wake up call for America,” Alexandr Wang, the CEO of Scale AI, commented on social media.

But at the same time, many Americans—including much of the tech industry—appear to be lauding this Chinese AI. As of this morning, DeepSeek had overtaken ChatGPT as the top free application on Apple’s mobile-app store in the U.S. Researchers, executives, and investors have been heaping on praise. The new DeepSeek model “is one of the most amazing and impressive breakthroughs I’ve ever seen,” the venture capitalist Marc Andreessen, an outspoken supporter of Trump, wrote on X. The program shows “the power of open research,” Yann LeCun, Meta’s chief AI scientist, wrote online.

Indeed, the most notable feature of DeepSeek may be not that it is Chinese, but that it is relatively open. Unlike top American AI labs—OpenAI, Anthropic, and Google DeepMind—which keep their research almost entirely under wraps, DeepSeek has made the program’s final code, as well as an in-depth technical explanation of the program, free to view, download, and modify. In other words, anybody from any country, including the U.S., can use, adapt, and even improve upon the program. That openness makes DeepSeek a boon for American start-ups and researchers—and an even bigger threat to the top U.S. companies, as well as the government’s national-security interests.

To understand what’s so impressive about DeepSeek, one has to look back to December, when OpenAI launched its own technical breakthrough: the full release of o1, a new kind of AI model that, unlike all the “GPT”-style programs before it, appears able to “reason” through challenging problems. o1 displayed leaps in performance on some of the most challenging math, coding, and other tests available, and sent the rest of the AI industry scrambling to replicate the new reasoning model—which OpenAI disclosed very few technical details about. The start-up, and thus the American AI industry, were on top. (The Atlantic recently entered into a corporate partnership with OpenAI.)

DeepSeek, less than two months later, not only exhibits those same “reasoning” capabilities apparently at much lower costs, but has spilled at least one way to match OpenAI’s more covert methods to the rest of the world. The program is not entirely open-source—its training data, for instance, and the fine details of its creation are not public—but, unlike with ChatGPT, Claude, or Gemini, researchers and start-ups can still study the DeepSearch research paper and directly work with its code. OpenAI has enormous amounts of capital, computer chips, and other resources, and has been working on AI for a decade. In comparison, DeepSeek is a smaller team formed two years ago with far less access to essential AI hardware, because of U.S. export controls on advanced AI chips, but it has relied on various software and efficiency improvements to catch up. DeepSeek has reported that the final training run of a previous iteration of the model that R1 is built from, released in December, cost less than $6 million. Meanwhile, Dario Amodei, the CEO of Anthropic, has said that U.S. companies are already spending on the order of $1 billion to train future models. Exactly how much the latest DeepSeek cost to build is uncertain—some researchers and executives, including Wang, have cast doubt on just how cheap it could have been—but the price for software developers to incorporate DeepSeek-R1 into their own products is roughly 95 percent cheaper than incorporating OpenAI’s o1, as measured by the price of every “token”—basically, every word—the model generates.

DeepSeek’s success has abruptly forced a wedge between Americans most directly invested in outcompeting China and those who benefit from any access to the best, most reliable AI models. (It’s a divide that echoes Americans’ attitudes about TikTok—China hawks versus content creators—and China’s other apps and platforms.) For the start-up and research community, DeepSeek is an enormous win. “A non-US company is keeping the original mission of OpenAI alive,” Jim Fan, a top AI researcher at the chipmaker Nvidia and former OpenAI employee, wrote on X. “Truly open, frontier research that empowers all.”

But for America’s top AI companies, and the nation’s government, what DeepSeek represents is unclear. The stocks of many major tech firms—including Nvidia, Alphabet, and Microsoft—dropped this morning amid the excitement around the Chinese model. And Meta, which has branded itself as a champion of open-source models in contrast to OpenAI, now seems a step behind. (The company is reportedly panicking.) To some investors, all those massive data centers, billions of dollars of investment, or even the half-a-trillion-dollar AI-infrastructure joint venture from OpenAI, Oracle, and SoftBank, which Trump recently announced from the White House, could seem far less essential. Maybe bigger AI isn’t better. For those who fear that AI will strengthen “the Chinese Communist Party’s global influence,” as OpenAI wrote in a recent lobbying document, this is legitimately concerning: The DeepSeek app refuses to answer questions about, for instance, the Tiananmen Square protests and massacre of 1989 (although the censorship may be relatively easy to circumvent).

None of that is to say the AI boom is over, or will take a radically different form going forward. The next iteration of OpenAI’s reasoning models, o3, appears far more powerful than o1 and will soon be available to the public. There are some signs that DeepSeek trained on ChatGPT outputs (outputting “I’m ChatGPT” when asked what model it is), although perhaps not intentionally—if that’s the case, it’s possible that DeepSeek could only get a head start thanks to other high-quality chatbots. America’s AI innovation is accelerating, and its major forms are beginning to take on a technical research focus other than reasoning: “agents,” or AI systems that can use computers on behalf of humans. American tech giants could, in the end, even benefit. Satya Nadella, the CEO of Microsoft, framed DeepSeek as a win: More efficient AI means that use of AI across the board will “skyrocket, turning it into a commodity we just can’t get enough of,” he wrote on X today—which, if true, would help Microsoft’s profits as well.

Still, the pressure is on OpenAI, Google, and their competitors to maintain their edge. With the release of DeepSeek, the nature of any U.S.-China AI “arms race” has shifted. Preventing AI computer chips and code from spreading to China evidently has not tamped the ability of researchers and companies located there to innovate. And the relatively transparent, publicly available version of DeepSeek, rather than leading American programs, could mean Chinese programs and approaches become global technological standards for AI—akin to how the open-source Linux operating system is now standard for major web servers and supercomputers. Being democratic—in the sense of vesting power in software developers and users—is precisely what has made DeepSeek a success. If Chinese AI maintains its transparency and accessibility, despite emerging from an authoritarian regime whose citizens can’t even freely use the web, it is moving in exactly the opposite direction of where America’s tech industry is heading.

During an American election, a rich man can hand out $1 million checks to prospective voters. Companies and people can use secretly funded “dark money” nonprofits to donate unlimited money, anonymously, to super PACs, which can then spend it on advertising campaigns. Pod­casters, partisans, or anyone, really, can tell outrageous, incendiary lies about a candidate. They can boost those falsehoods through targeted online advertising. No special courts or election rules can stop the disinformation from spreading before voters see it. The court of public opinion, which over the past decade has seen and heard everything, no longer cares. U.S. elections are now a political Las Vegas: Anything goes.

But that’s not the way elections are run in other countries. In Britain, political parties are, at least during the run-up to an election, limited to spending no more than £54,010 per candidate. In Germany, as in many other European countries, the state funds political parties, proportionate to their number of elected parliamentarians, so that politicians do not have to depend on, and become corrupted by, wealthy donors. In Poland, courts fast-track election-­related libel cases in the weeks before a vote in order to discourage people from lying.

Nor is this unique to Europe. Many democracies have state or public media that are obligated, at least in principle, to give equal time to all sides. Many require political donations to be transparent, with the names of donors listed in an online registry. Many have limits on political advertising. Some countries also have rules about hate speech and indict people who break them.

Countries apply these laws to create conditions for fair debate, to build trust in the system, and to inspire confidence in the winning candidates. Some democracies believe that transparency matters—­that voters should know who is funding their candidates, as well as who is paying for political messages on social media or anywhere else. In some places, these rules have a loftier goal: to prevent the rise of anti­democratic extremism of the kind that has engulfed democracies—­and especially European democracies—­­in the past.

But for how much longer can democracies pursue these goals? We live in a world in which algorithms controlled by American and Chinese oligarchs choose the messages and images seen by millions of people; in which money can move through secret bank accounts with the help of crypto schemes; and in which this dark money can then boost anonymous social-media accounts with the aim of shaping public opinion. In such a world, how can any election rules be enforced? If you are Albania, or even the United Kingdom, do you still get to set the parameters of your public debate? Or are you now forced to be Las Vegas too?

Although it’s easy to get distracted by the schoolyard nicknames and irresponsible pedophilia accusations that Elon Musk flings around, these are the real questions posed by his open, aggressive use of X to spread false information and promote extremist and anti-European politicians in the U.K., Germany, and elsewhere. The integrity of elections—­and the possibility of debate untainted by misinformation injected from abroad—is equally challenged by TikTok, the Chinese platform, and by Mark Zuckerberg’s Meta, whose subsidiaries include Facebook, Instagram, WhatsApp, and Threads. TikTok says the company does not accept any paid political advertising. Meta, which announced in January that it is abandoning fact-checking on its sites in the U.S., also says it will continue to comply with European laws. But even before Zucker­berg’s radical policy change, these promises were empty. Meta’s vaunted content curation and moderation have never been transparent. Nobody knew, and nobody knows, what exactly Facebook’s algorithm was promoting and why. Even an occasional user of these platforms encounters spammers, scammers, and opaque accounts running foreign influence operations. No guide to the algorithm, and no real choices about it, are available on Meta products, X, or TikTok.

In truth, no one knows if any platforms really comply with political-funding rules either, because nobody outside the companies can fully monitor what happens online during an intense election campaign—and after the voting has ended, it’s too late. According to declassified Romanian-intelligence documents, someone allegedly spent more than $1 million on TikTok content in the 18 months before an election in support of a Romanian presidential candidate who declared that he himself had spent nothing at all. In a belated attempt to address this and other alleged discrepancies, a Romanian court canceled the first round of that election, a decision that itself damaged Romanian democracy.

Not all of this is new. Surreptitious political-party funding was a feature of the Cold War, and the Russian government has continued this practice, sometimes by offering deals to foreign business­people close to pro-Russian politicians. Press moguls with international political ambitions are hardly a novelty. Rupert Murdoch, an Australian who has U.S. citizenship, has long played an outsize role in U.K. politics through his media companies. John Major, the former British prime minister and Conservative Party leader, has said that in 1997, Murdoch threatened to pull his newspapers’ support unless the prime minister pursued a more anti-­European policy. Major refused. Murdoch has said, “I have never asked a prime minister for anything,” but one of his Conservative-­leaning tabloids, The Sun, did endorse the Labour Party in the next election. Major lost.

That incident now seems almost quaint. Even at the height of its influence, the print edition of The Sun sold 4 million copies a day. More to the point, it operated, and still does, within the constraints of U.K. rules and regulations, as do all broadcast and print media. Murdoch’s newspapers take British libel and hate-speech laws into consideration when they run stories. His business strategy is necessarily shaped by rules limiting what a single company can own. After his journalists were accused of hacking phones and bribing police in the early 2000s, Murdoch himself had to testify before an investigative commission, and he closed down one of his tabloids for good.

[McKay Coppins: Europe braces for Trump]

Social media not only has far greater reach—Musk’s personal X account has more than 212 million followers, giving him enormous power to set the news agenda around the world—it also exists outside the legal system. Under the American law known as Section 230, passed nearly three decades ago, internet platforms are not treated as publishers in the U.S. In practice, neither Facebook nor X has the same legal responsibility for what appears on their platforms as do, say, The Wall Street Journal and CNN. And this, too, has consequences: Americans have created the information climate that other countries must accept, and this allows deceptive election practices to thrive. If countries don’t have their own laws, and until recently most did not, Section 230 effectively requires them to treat social-media companies as if they exist outside their legal systems too.

Brazil broke with this pattern last year, when a judge demanded that Musk comply with Brazilian laws against spreading misinformation and political extremism, and forced X offline until he did. Several European countries, including the U.K., Germany, and France, have also passed laws designed to bring the platforms into compliance with their own legal systems, mandating fines for companies that violate hate-speech laws or host other illegal content. But these laws are controversial and hard to enforce. Besides, “illegal speech” is not necessarily the central problem. No laws prevented Musk from interviewing Alice Weidel, a leader of the far-right Alternative for Germany (AfD) party, on X, thereby providing her with a huge platform, available to no other political candidate, in the month before a national election. The interview, which included several glaringly false statements (among others, that Weidel was the “leading” candidate), was viewed 45 million times in 24 hours, a number far beyond the reach of any German public or private media.

Only one institution on the planet is large enough and powerful enough to write and enforce laws that could make the tech companies change their policies. Partly for that reason, the European Union may soon become one of the Trump administration’s most prominent targets. In theory, the EU’s Digital Services Act, which took full effect last year, can be used to regulate, fine, and, in extreme circumstances, ban internet companies whose practices clash with European laws. Yet a primary intent of the act is not punitive, but rather to open up the platforms: to allow vetted researchers access to platform data, and to give citizens more transparency about what they hear and see. Freedom of speech also means the right to receive information, and at the moment social-media companies operate behind a curtain. We don’t know if they are promoting or suppressing certain points of view, curbing or encouraging orchestrated political campaigns, discouraging or provoking violent riots. Above all, we don’t know who is paying for misinformation to be spread online.

In the past, the EU has not hesitated to try to apply European law to tech companies. Over the past decade, for example, Google has faced three fines totaling more than $8 billion for breaking antitrust law (though one of these fines was overturned by the EU’s General Court in 2024).

In November, the European Commission fined Meta more than $800 million for unfair trade practices. But for how much longer will the EU have this authority? In the fall, J. D. Vance issued an extraordinarily unsubtle threat, one that is frequently repeated in Europe. “If NATO wants us to continue supporting them and NATO wants us to continue to be a good participant in this military alliance,” Vance told an interviewer, “why don’t you respect American values and respect free speech?” Mark Zuckerberg, echoing Vance’s misuse of the expression free speech to mean “freedom to conceal company practices from the public,” put it even more crudely. In a conversation with Joe Rogan in January, Zuckerberg said he feels “optimistic” that President Donald Trump will intervene to stop the EU from enforcing its own antitrust laws: “I think he just wants America to win.”

Does America “winning” mean that European democracies, and maybe other democracies, lose? Some European politicians think it might. Robert Habeck, the German vice chancellor and a leader of that country’s Green Party, believes that Musk’s frenzies of political activity on X aren’t the random blurts of an addled mind, but rather are “logical and systematic.” In his New Year’s address, Habeck said that Musk is deliberately “strengthening those who are weakening Europe,” including the explicitly anti-European AfD. This, he believes, is because “a weak Europe is in the interest of those for whom regulation is an inappropriate limitation of their power.”

Until recently, Russia was the most important state seeking to undermine European institutions. Vladimir Putin has long disliked the EU because it restricts Russian companies’ ability to intimidate and bribe European political leaders and companies, and because the EU is larger and more powerful than Russia, whereas European countries on their own are not. Now a group of American oligarchs also want to undermine European institutions, because they don’t want to be regulated—and they may have the American president on their side. Quite soon, the European Union, along with Great Britain and other democracies around the world, might find that they have to choose between their alliance with the United States and their ability to run their own elections and select their own leaders without the pressure of aggressive outside manipulation. Ironically, countries, such as Brazil, that don’t have the same deep military, economic, and cultural ties to the U.S. may find it easier to maintain the sovereignty of their political systems and the transparency of their information ecosystems than Europeans.

A crunch point is imminent, when the European Commission finally concludes a year-long investigation into X. Tellingly, two people who have advised the commission on this investigation would talk with me only off the record, because the potential for reprisals against them and their organizations—­whether it be online trolling and harassment or lawsuits—­is too great. Still, both advisers said that the commission has the power to protect Europe’s sovereignty, and to force the platforms to be more transparent. “The commission should look at the raft of laws and rules it has available and see how they can be applied,” one of them told me, “always remembering that this is not about taking action against a person’s voice. This is the commission saying that everyone’s voice should be equal.”

At least in theory, no country is obligated to become an electoral Las Vegas, as America has. Global democracies could demand greater transparency around the use of algorithms, both on social media and in the online-advertising market more broadly. They could offer consumers more control over what they see, and more information about what they don’t see. They could enforce their own campaign-funding laws. These changes could make the internet more open and fair, and therefore a better, safer place for the exercise of free speech. If the chances of success seem narrow, it’s not because of the lack of a viable legal framework—­rather it’s because, at the moment, cowardice is as viral as one of Musk’s tweets.

This article appears in the March 2025 print edition with the headline “Can Europe Stop Elon Musk?”