Demis Hassabis wants America to police the world’s AI. Xi Jinping will offer a rival answer on Friday.
China's mooted shift on open-source AI cannot survive American safety testing.
Demis Hassabis, Google DeepMind co-founder and one of the individuals most directly responsible for where we are right now, technically and philosophically both, has a plan.
That plan is America.
On Tuesday, Hassabis published a new manifesto calling for a frontier AI standards body on Substack. (Yes, really, Axios confirms it’s him.)
“By safely stewarding AGI into the world, we can enter a new golden age of scientific discovery and progress, and usher in a bright future of incredible human flourishing,” he said.
“We”, as in humanity? “We” as in some sort of elite, perhaps his fellow Nobel laureates?
Nope.
“The US is well positioned, given its economic and technical standing, to take the first step in developing such a framework,” he also wrote.
Perhaps this “US-initiated effort” would get the rest of the world to get its act together, Hassabis muses, but he’s clearly not bothered about that potential, far-off future. Right here, right now, the US should be policing the world’s AI.
This will, of course, play incredibly well with the Trump administration.
It is going to play a helluva lot less well with China.
China, unsurprisingly, has a different plan
Hassabis’s “USA! USA! USA!” comes as the United Nations moves, with unusual haste, towards global AI regulation. That is probably no coincidence. If America is to dominate frontier AI (and eventually AGI) regulation, it has to move now.
But more intriguingly, Hassabis floated his plan three days before Xi Jinping headlines the 2026 World AI Conference and High-Level Meeting on Global AI Governance in Shanghai.
