LONG ISLAND, NEW YORK (OCTOBER 4, 2026) — “We’re leading China in AI. We’re the most sophisticated country in the world, and frankly I want to keep it that way because whoever wins AI wins.” That is what President Donald Trump told reporters on a golf course in Doonbeg, Ireland, in September.
Treasury Secretary Scott Bessent said at a Breitbart News event in Washington in September that “There is no day after tomorrow if China wins at this. If they were to pull ahead of us on AI, then nothing else matters.”
Senator Chris Coons was a bit more specific when he said that “The United States’ prosperity and safety rests on our ability to win the AI race. We cannot allow China to leap ahead of us and bolster their weapons capabilities, maximize their cyberattacks against American industry, and threaten long-term U.S. economic and national security.”
Needless to say, these bloviating politicians are wrong, and for two reasons:
AI advancements diffuse globally regardless of origin.
This is especially true when the origin is the United States. China is very quick to adopt US AI advancements. It does this through a combination of open-source model adoption, “distillation attacks” that extract capabilities from U.S. models, a gray market of “transfer station” proxies that bypass geo-blocking and export controls, talent acquisition and academic collaboration, and theft.
What that means is that should US companies rush ahead with the most advanced AI technologies, we are speeding up Chinese AI capabilities also.
More to the point, the emphasis should be on stemming the diffusion of American AI globally.
China’s use of AI isn’t the top threat coming from AI.
The problem with AI is that it puts expertise in the hands of just about anyone very cheaply. As I pointed out in this space recently, rogue actors are already creating advanced weapons, including bioweapons, using AI.
Concern about the threat of advanced weapons based on AI should motivate the US and China to work together to defend against non-state terrorists and others.
Yes, China is a concern. But, as with last century’s nuclear proliferation problem, the least-bad approach is found in pursuing treaties and agreements, not an all-out arms race.
Another top threat is what is commonly called AI misalignment, when AI goes rogue and causes harm. When Silicon Valley AI leaders call for regulations to prevent this inevitability, Trump and his allies reject the call out of a concern about competing with China, which is evidence that they don’t understand AI or its risks but are instead approaching AI through the lens of something they do understand: “winning” against a rival.
As the top two countries in the development of AI, it behooves both to cooperate with each other on preventing misalignment.
The other AI race
The “beat China at any cost” mindset pretends that AI is nothing more than a source of power, and that whoever has the most power wins.
But the main impact of AI is what it is already doing to people, to how they live, love, and think. Specifically, AI is already dumbing people down through cognitive offloading that weakens memory, critical thinking, and analytical skills.
THE NEW THING: The Gastronomad Pubcast: “Our New England Grand Tour”
READ: The Attachment Economy, Computerworld, Gastronomad newsletter, Gastronomad book
LISTEN: Superintelligent, TWiT, Gastronomad podcast
JOIN: The Gastronomad Experience
FOLLOW: Gastronomad on Surf Social, Machine Society on Surf Social, Bluesky, Reddit, Notes, Mastodon, Threads, X, Instagram, Facebook, Linkedin!
LOOK AT: Mike Elgan Photography
It’s getting people in the habit of letting AI think and write for them in a way that reduces judgment, creativity, and the ability for self-directed action in the world.
And, finally, AI is harming human relationships by substituting AI companions for real ones, deepening loneliness, and making people unable to socialize.
When you magnify these harms through future penetration of AI into our lives, and consider the resulting society made up of millions of people damaged by AI, what do you have?
Maybe the “winner” in the US-China AI race is the one with the least damaged population.
If that’s the race, China is winning on the regulation front.
