Skip to content

US and China urged to unite on AI safety before Trump-Xi summit

A new AI arms race demands diplomacy—not division. Could the Trump-Xi summit defuse the ticking time bomb of unchecked artificial intelligence?

The image shows an organizational chart of the United States Air Force, with the Secretary of...
The image shows an organizational chart of the United States Air Force, with the Secretary of Defense at the top. The chart is composed of several boxes connected by arrows, each box representing a different branch of the organization. The text written on the chart provides further details about the roles and responsibilities of each branch.

US and China urged to unite on AI safety before Trump-Xi summit

Calls for stronger cooperation between the US and China on AI safety have grown louder after the release of Anthropic’s Claude Mythos. Security experts warn that unchecked AI development could pose serious risks, including cyber threats with global consequences. A proposed government-to-government dialogue aims to address these concerns ahead of the upcoming Trump-Xi summit in Beijing. The push for a US-China AI risk dialogue follows concerns over advanced models like Claude Mythos. Experts, including Thomas Friedman, Sebastian Mallaby, and Angela Huyue Zhang, argue that both nations must work together to control potential dangers. While not aimed at formal agreements, the talks would focus on sharing safety guidelines and understanding common risks.

China has already updated its AI regulations, signalling readiness for discussions. Meanwhile, the US is considering stricter reviews for new AI models. A recent risk assessment of China’s open-source model Kimi K2.5 highlighted higher vulnerabilities compared to US-developed systems. The proposed dialogue would establish official channels for reporting AI safety incidents and misuse. Experts also warn that a cyberattack on China’s manufacturing sector could disrupt global supply chains, much like the COVID-19 pandemic. Recognised countermeasures, such as alignment techniques and controlled AI environments, would be key topics.

The Trump-Xi summit in Beijing presents an opportunity to launch formal AI risk discussions. Both nations would exchange nonbinding safety guidelines and limited incident data. The goal is to build a shared understanding of AI’s benefits and hazards without negotiating specific policies.

Read also:

Latest