The summit between President Trump and President Xi Jinping in Beijing next week could be the most significant encounter between American and Chinese leaders since Richard Nixon met Mao Zedong in Beijing in 1972.
That summit eased decades of Sino-American animosity and forged a tacit alliance between the United States and China against the Soviet Union. This summit comes at a similar transformational moment in world affairs, when there is a new shared threat to both China and America. It is a metastasizing disorder that could destabilize the world and harm both countries unless they figure out a way to simultaneously compete and collaborate against a growing list of challenges. These challenges can be successfully confronted only by their collective action — starting with the United States and China together creating guardrails against the malign uses of A.I., now that the latest models have demonstrated staggeringly powerful cyberattack capabilities.
Two paradigm shifts have changed the world since the Nixon-Mao summit. The first — still not widely appreciated, although the alarm bells are now ringing off the wall — is the emergence of these new, asymmetric artificial intelligence tools that could superempower small, malign actors, be they terrorists, anarchists, criminals, political groups or small nation-states.
Two guys in a cave with a laptop, access to the latest A.I. models and a Starlink terminal could attack the critical infrastructure of any society.
The second has to do with globalization. The Nixon-Mao summit began the process of taking the world from disconnected to much more connected and then interconnected. When Nixon and Mao began easing China out of its isolation from the global economy — which Deng Xiaoping then vastly accelerated by shifting China to state-led capitalism — they unleashed a cascade of economic and technological forces.
By the time the early 21st century rolled around, the combination of China joining the World Trade Organization and the world being wired with the internet meant that more people in more places could compete, connect and collaborate in more ways for less money on more things than at any other time in human history. It is why I wrote a book in 2005 titled “The World Is Flat.”
It is in the nature of technological change, though, that each major step forward comes faster than the previous one, because it builds on the tools that the previous era unleashed. So, years after I argued that the world is flat, technology, and other forces, marched on and took us, as Dov Seidman, the founder of The HOW Institute for Society, argued, from interconnected to interdependent, or as he puts it, from flat to “fused.”
You could unplug from the flat world. There is no escaping the fused world. We are all going to rise and fall together now.
That is not only because advances in the internet, smartphones, fiber optics, satellites and wireless communication have fused us technologically more than ever before. It is also because a set of planetary-scale challenges have fused our fates together more than ever before as well. These challenges are so large in scope and so indifferent to national borders that no single state, however powerful, can address or escape them alone.
We know what they are: mitigating climate change, preventing the spread of nuclear and biological weapons, managing global migrations, controlling pandemics, keeping global supply chains that we all now depend on operating smoothly and — most important and immediate — managing this new A.I. species we have conjured up.
We have been able to postpone or get by with limited collaboration on many of these planetary-scale issues, but time is up on A.I.’s cyberattacking powers. There is no kicking this can down the road. There is no more road.
For years, notes Craig Mundie — a former head of research and strategy at Microsoft and my tutor and partner in thinking about this new A.I. threat — the United States and China have regularly poked and probed each other, and have embedded malware infrastructure and stolen information from each other with covert cyberoperations. But they also knew, Mundie noted, that if China took out our electricity grids, we could take out theirs, and that if they could turn the lights out in Washington, we could do the same in Beijing. It’s the same as with nuclear weapons: “They had recreated mutually assured destruction,” Mundie said.
But now guess who’s coming to dinner? A new set of actors, potentially very dangerous, and they’re not just countries. Yet they can threaten us both.
These are the agentic A.I. systems recently disclosed by Anthropic and OpenAI that could give tools to small cyberattackers to disrupt both China’s economy and ours — and anybody else’s — with very little money and virtually no expertise. You can bet that other U.S. models, like Google’s Gemini, and soon China’s A.I. models, will offer the same powers.
Because companies in the United States and China have been the first to produce these agentic systems, “the two of them must take the lead in controlling their distribution and building defenses to protect themselves — and everyone else if they leak out,” Mundie said.
Anthropic and OpenAI say their newest models are so potent at finding and exploiting flaws in software that both companies have chosen to limit their distribution for now. But it is only a matter of time before they escape into the wild, if they haven’t already.
“This should be a big motivator for the two countries to come together — if only on this narrow issue, which is now a clear and present danger to both,” Mundie argues.
It is not asking the impossible. China and the United States were able to cooperate in the days of Nixon and Mao, Mundie concluded, “because we had a common problem, the Soviet Union. Well, now we have another common problem. It is not another country, it’s a technology — the emerging risks from asymmetric cyberthreats from agentic A.I. systems.”
The old G2, the United States and China, need to work with what what I’ll call the new I7 — Anthropic, Google/Alphabet, OpenAI, Meta, Alibaba Group, DeepSeek and ByteDance — to figure out a way to get the best from these new A.I. models while cushioning against the worst. The governments can’t solve this by themselves, and neither can the companies.
In a development that received too little attention because of the Iran war, Trump is reportedly now considering imposing oversight on A.I. models before they are made publicly available. That is very wise of Trump. People need to wake up: We are entering a world in which private companies can now, in effect, split the atom, in terms of the power they can unleash in every direction.
“And as with splitting the atom, you can either make electricity, or bombs,” Mundie said. The same is true with agentic A.I. “We have the power to do unlimited good or create weapons — hugely asymmetric weapons.”
The topic of agentic A.I. is expected to be on the Trump-Xi agenda. What would truly make this the most significant U.S.-China summit since Mao and Nixon is not just that the two men talk about it, but that they decide to work together on it — now. Later will be too late. It’s just coming too fast.
Even if many leaders in Washington, Beijing and, for that matter, Moscow still haven’t grasped it, this is the first era of human history in which we Homo sapiens must govern, innovate, collaborate and coexist at a planetary scale to thrive. We will either build complex, adaptive coalitions to do that or we will be overwhelmed together.
Our fates are now fused.
The Times is committed to publishing a diversity of letters to the editor. We’d like to hear what you think about this or any of our articles. Here are some tips. And here’s our email: letters@nytimes.com.
Follow the New York Times Opinion section on Facebook, Instagram, TikTok, Bluesky, WhatsApp and Threads.
