US and China bare their teeth over AI: from fear of the apocalypse to a tech cold war

An entire generation of Americans was marked by the terror they felt when a Soviet device flew over their heads: the Sputnik probe, the planet’s first artificial satellite. That “Sputnik moment” not only spurred the United States in the space race — which ended with the USSR derailed and Neil Armstrong stepping onto the Moon — it also became the great metaphor for the flashiest contest of the Cold War. These days, the vast rivalry between the two superpowers of our era has a different actor, China instead of the U.S.S.R., but it is still being waged on every front: trade, space, weapons, and also artificial intelligence.

The arrival of far more powerful models has revolutionized the economy and triggered a major geopolitical clash over who will control a technology that is already shaping the present and will redefine the future. And now that the greatest fears about the danger of artificial intelligence (AI) have begun to materialize in the global public debate, the two powers have collided again. “Whoever wins AI wins,” Donald Trump summed up on Sunday, and on Monday he pointed directly at the enemy, claiming there is a “SICK conspiracy going on against AI and Data Centers, and the only one that is happy about it is China.”

Beijing responded with indignation, accusing the U.S. of reverting to Cold War rhetoric. Rarely have the two powers so nakedly shown that AI regulation will define their global rivalry. On September 24, these tensions will be concentrated in a single point on the planet — the few square meters of the Oval Office — where Trump will receive his counterpart Xi Jinping.

In this vast conflict over AI power there are numerous decisive battles, from control of the supply chain for the most powerful chips (which the U.S. denies to China) to the mass of engineers able to develop the best tools. But these last few days the AI race has focused on a new front: how to prevent the apocalypse. Exaggerated or not, the words of Jacob Coxon, the Anthropic researcher who quit his job, have accelerated matters. “The people building AI earnestly believe that it could kill us all by the end of the decade,” Coxon posted on social media on Wednesday, and by Monday we were already seeing China’s anger at how the doomsday rhetoric had affected Beijing. Suddenly, the main argument against slowing a potentially dangerous technology is no longer business, but fear that China will win it.

The U.S. president dismissed out of hand the cascade of tech leaders calling to slow AI development with a single argument: the need to maintain the lead over the Asian rival challenging American power. The White House occupant is not the least fazed by warnings from industry leaders, such as the scenario sketched by Dario Amodei, CEO of Anthropic, in which a swarm of AI agents ravages the web: “Given the accelerating rate of AI capability development, it’s my worry that in 6–12 months such a swarm could be capable of taking over the entire internet with a persistent botnet (potentially causing hundreds of billions of dollars in damage), and that the scale of damage would continue to increase from there if AI becomes more powerful without the necessary guardrails,” he wrote in a lengthy essay.

Curiously, Anthropic’s co-founder and CEO has been rebuked by both China and the U.S. Speaking from Beijing, Foreign Ministry spokesman Guo Jiakun replied that “alarmism, confrontation and ruthless competition will only disrupt the global governance process for AI, which serves no one’s interests.” Amodei said that “I agree with Secretary Bessent that a Chinese lead in AI would pose grave danger for the United States and the world,” and urged continued restrictions on sales of cutting-edge chips to the Asian rival. But Washington also lashed out at Amodei.

What Amodei proposed, and what so irritates Trump, is slowing the development of new AI models through external “embedded evaluators” who could inspect these programs before they are launched. In an AI market where capabilities can double in just weeks, having to endure audits that slow product rollouts would give China the chance to overtake U.S. companies, which it is already closing in on. And the White House is not willing to experience its own Sputnik Moment with a Chinese superintelligence sweeping aside U.S. competitors. It is not just about who conquers the chatbot market — it could be an existential matter in geopolitical terms.

A Chinese competitor to ChatGPT

The U.S. side already got a taste of it in January 2025 when a small, little-known Chinese company called DeepSeek, launched a strong competitor to ChatGPT: a tool that worked as well as or better than the rest. It was also free and open source, and had been developed in China despite the chip embargo and using far fewer resources than its rivals. While U.S. companies were competing by spending billions, DeepSeek had found a computational shortcut that allowed it to go as far without such an investment. Accusations that the Chinese were copying U.S. technology were quick to emerge. In the future, the risk is that Chinese technology may upend the global board. Right now, the danger is that Chinese companies could steal market share from U.S. firms with cheaper chatbots for the public.

In that context, OpenAI, Anthropic and Google appear to agree on self-regulation. Silicon Valley outlet The Information reported that the three companies have been discussing that idea for some time: imposing rules on themselves that will allow them to advance cautiously but without too many hindrances. The fact that their leaders are aligning on the public message seems to bear that out: Amodei — regarded as the most reflective among the AI magnates — first published his proposal, followed by Demis Hassabis, Google’s chief scientist, and Sam Altman, OpenAI’s CEO, who advocated having “consistent rules to manage frontier risk [the most advanced AI] so that we can maximize the benefits.” If the major U.S. tech companies manage to impose a regulatory standard in their market, they could clip the wings of Asian competitors.

The closest example is TikTok: on national security grounds, Washington barred Chinese company ByteDance from continuing to operate on U.S. soil unless it sold its popular short-video social network to local companies. After months of disagreement, the situation was resolved with a phone call between Trump and Xi. In that call, held in September 2025, the two leaders arranged TikTok’s future and also agreed to meet twice this year: the Beijing summit held in May and the one taking place next week in Washington. The summit could not come at a more heated moment for AI’s future and, in fact, both countries are already preparing to discuss the safety of the most powerful models. When they spoke a year ago, the two leaders surely knew they would keep clashing in the future. What they perhaps did not imagine is that on September 24 they would have to address how to stop all the apocalypses foretold by tech gurus.

Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition