The US-China AI Arms Race Could Be Heading Towards a ‘Chernobyl Moment’

I met with China's top AI experts.

They are freaking out, too.

Last month I attended a major AI conference

in Beijing's bustling high-tech district.

It featured fascinating demonstrations

of the latest humanoid robots

and sessions on self-improving models

that tweak their own code and advance indefinitely.

But I came away with one key takeaway:

that the US and China should put their fierce AI rivalry

to one side and work together.

Let me explain why.

The cybersecurity and systemic risks associated

with frontier AI are simply too important to ignore.

The US largely views China's AI advances as a threat

to both its economic prospects and national security,

imposing tight restrictions on chips

and chip-making equipment.

The US recently ordered Anthropic

to prevent foreign nationals

from using its most powerful models, Mythos and Fable 5.

As WIRED reported, a South Korean telecom giant

with alleged links to China was a particular concern.

In response, Anthropic revoked access to everyone.

The conference I attended reinforced the idea

that both the US and China,

and potentially everyone, stands to lose

if AI is developed too quickly and recklessly.

As the world's two AI superpowers,

they're responsible for the most advanced models

and cooperation could help mitigate risks associated

with powerful agentic systems conducting cyber attacks

or failing in catastrophic ways.

One expert I spoke to after the conference

likened the current situation

to the way the US and the Soviet Union

were eventually forced to work together on nuclear safety.

In that case, the benefits of collaboration

ultimately outweighed the national security risks.

He told me, One thing almost everyone in AI

can agree on right now

is that AI doesn't need a Chernobyl moment.

For more like this, subscribe

to my newsletter AI Lab at wired.com/newsletters.