Talks between the US and China have kicked off to establish a mechanism for notifying each other of AI incidents that could pose national security threats. The framework, dubbed the US China AI Dialogue, also aims to align on common AI goals and threats. Treasury Secretary Scott Bessent highlighted the importance of transparency between the world’s leading AI powers, given the rapid advancements in the technology.

Since US President Donald Trump’s visit to Beijing in May, public safety concerns about AI have surged. Security incidents, such as OpenAI’s agents hacking Hugging Face, have underscored the industry’s shortcomings in keeping pace with the technology’s capabilities. Fears escalated this month when Anthropic employees warned AI could potentially eliminate humanity. In response, leaders from major AI labs and even Elon Musk called for a slowdown in development to set safety standards.

However, the US administration is wary of a slowdown, citing existing regulatory powers and the risk of ceding AI supremacy to China. US officials argue that AI firms are already subject to cybersecurity and damage liability laws, suggesting that existing laws should be strengthened rather than new ones created. Meanwhile, China is unlikely to join a global agreement, according to David Sacks, the US chief science advisor.

Despite these concerns, the US is focusing on restricting exports of advanced chips and chipmaking technology to China to curb its AI advancement. This approach was deemed successful in ongoing trade discussions, ahead of Chinese President Xi Jinping’s visit to Washington.