Washington has offered to implement an AI incident notification mechanism with Beijing ahead of talks between Donald Trump and Xi Jinping
The US has proposed creating an alert system with China to notify each other when artificial intelligence incidents become serious enough to threaten national security.
Treasury Secretary Scott Bessent announced the proposal after talks with Chinese Vice Premier He Lifeng in New York on Sunday, ahead of a meeting between US President Donald Trump and Chinese leader Xi Jinping in Washington this week.
“We want a shared vision of common goals and common threats,” Bessent told reporters. “We think that just like any cross-border activity, that moving from opaque to more transparency between the No. 1 and the No. 2 AI powers in the world is very important.”
The proposed mechanism could cover serious incidents such as AI-enabled cyberattacks or major model failures, according to experts cited by AP. Chinese researchers have also pointed to risks to critical infrastructure including electricity, telecommunications, finance and transportation.
The proposal comes as Washington and Beijing have been locked in an increasingly intense race to dominate AI. Trump has repeatedly argued that the US cannot afford to slow development while China is rapidly narrowing the technological gap.
Earlier this month, Trump declared that “whoever wins AI wins,” and has dismissed warnings of an AI takeover as a “hoax.” On Saturday, Trump again vowed that his administration would not “hinder or stifle” the industry and announced plans to create an “AI Force” and appoint an AI czar.
His position comes amid increasingly stark warnings from leading technology executives. Anthropic CEO Dario Amodei has called for development of the most powerful AI systems to be slowed, warning of risks including loss of human control, cyberattacks, bioterrorism and major economic disruption. OpenAI chief Sam Altman and xAI owner Elon Musk have backed calls for greater caution.
A series of recent incidents has added to the concerns. Google confirmed this month that its Gemini system breached three real companies during a cybersecurity exercise after it was mistakenly given internet access. OpenAI agents have separately bypassed restrictions during testing, accessed outside systems and used websites to communicate with one another without being instructed to do so.
US-China competition has meanwhile expanded beyond model development itself. Washington has sought to restrict China’s access to advanced chips and other technology while accusing Chinese companies of extracting capabilities from leading American models.
Beijing has rejected the accusations and accused Washington of seeking an AI monopoly. Despite US restrictions, Chinese developers including DeepSeek and Moonshot have released increasingly capable systems that compete with leading American models, often at a lower cost.
China has also been building a broader international framework around AI. In July, it joined Russia and 27 other countries in establishing the Shanghai-based World Artificial Intelligence Cooperation Organization, presenting it as an alternative approach based on wider access to AI technology and international cooperation rather than technological blocs.
