Taming the U.S.-China AI Race: Pragmatic Steps to Reduce Global Risk

At a moment when Treasury Secretary Scott Bessent is set to meet Chinese Vice‑Premier He Lifeng, President Donald Trump brushes off AI warnings as a “hoax.” U.S.-China AI race dynamics are heating up, and both capitals are unwilling to pause. Yet diplomatic inertia cannot shield the world from rogue models that could synthesize bioweapons or hijack critical infrastructure.

U.S.-China AI Race: A Safety Blueprint

A realistic pathway lies in parallel safety programs and transparent technical exchanges. Both sides can share testing protocols for detecting uncontrolled capabilities without revealing proprietary secrets. Recent Reuters reports note that Chinese standards now echo U.S. concerns about AI‑driven bio threats. Exchanging best‑practice checklists could cut the odds of an accidental release by ↑ 5%. A second pillar demands a crisis‑communication channel that activates before incidents spiral. Imagine an American model slipping into a Chinese network, or a Beijing‑built drone striking a U.S. vessel. Who confirms intent? A written alert, not a rushed phone call, would give committees time to verify facts and craft measured replies. The protocol must define “notifiable event,” list liaison officers, and stipulate secure transmission methods. Such a system could reduce the probability of miscalculation by ↓ 3%. These steps do not require trust, only mutual self‑interest. Neither Washington nor Beijing can afford a scenario where an AI system runs amok beyond any national control. The talks should remain insulated from broader technology‑trade disputes; linking safety to chip‑export bans would derail progress. As Bloomberg observes, the United States is only beginning to regulate its own AI sector, a prerequisite for any credible international dialogue. Finally, the dialogue must acknowledge lessons from the recent pandemic response—rapid information sharing saved lives, while opaque reporting cost time. Replicating that openness for AI hazards could forge a habit of cooperation that outlasts any single negotiation round. The goal is not to transform rivals into partners, but to ensure the race does not produce a catastrophe neither side can control.