US and China Seek AI Safety Dialogue Despite Chip Rivalry
Officials propose a notification system for AI incidents as export controls remain a major barrier to cooperation.
Key points
- US and China officials proposed a notification mechanism for AI incidents affecting national security.
- Both nations cite risks to critical infrastructure as a key area for potential cooperation.
- U.S. export controls on advanced chips remain a major barrier to broader AI collaboration.
U.S. President Donald Trump and Chinese leader Xi Jinping are set to meet this week to discuss the shared risks of artificial intelligence. The conversation comes as both nations race to dominate the sector while acknowledging that uncontrolled AI could pose serious threats to global security.
Despite intense geopolitical competition, officials from both sides have proposed a new mechanism for sharing information about AI-related emergencies. This includes incidents involving cyberattacks or loss of human control over advanced systems, areas where both countries see a need for coordinated response.
Shared risks drive the proposed dialogue
U.S. Treasury Secretary Scott Bessent confirmed that a notification framework is under discussion to handle AI incidents that rise to the level of national security. Experts note that both governments are increasingly concerned about how AI could be used to attack critical infrastructure such as power grids and telecommunications networks.
According to WTOP News, this shared concern offers a narrow but significant area for cooperation. While the broader relationship remains strained, the potential for technical confidence-building through emergency communication channels is viewed by some analysts as a pragmatic step forward.
Export controls remain the central obstacle
The primary friction point remains U.S. restrictions on the export of advanced chips to China. Beijing accuses Washington of seeking an AI monopoly, while the U.S. insists on maintaining a technological edge. Analysts suggest that any easing of these controls is unlikely during the current talks.
Narrow path for future collaboration
The trade-off for the U.S. is accepting limited engagement with a strategic rival to mitigate existential risks. The catch is that without resolving the underlying competitive tensions over technology access, the scope for meaningful long-term cooperation on AI safety remains quite restricted.
