The United States and China are preparing to expand their cooperation on artificial intelligence security, despite the technological rivalry between the two countries. Washington and Beijing have agreed to establish a direct communication channel for the prompt reporting of serious incidents involving advanced AI systems. The new mechanism will be one of the topics of discussion between Donald Trump and Xi Jinping, scheduled for 18–19 November in Shenzhen, China.
- The US and China have agreed to set up a channel for reporting dangerous incidents involving AI.
- The countries plan to share best practices on the verification and testing of cutting-edge AI models.
- This new dialogue will continue during the meeting between Trump and Xi on 18–19 November.
- Washington insists on maintaining human control over powerful AI systems.
Washington and Beijing are establishing a communication channel on artificial intelligence
US Secretary of the Treasury Scott Bessent He reported that the two countries had agreed to establish a special mechanism for the prompt reporting of dangerous incidents involving advanced artificial intelligence systems.
Its purpose is to prevent misunderstandings between Washington and Beijing in cases where the behaviour of AI could pose a threat to national security or trigger an international crisis.
Agreements regarding this channel were reached as early as the September meeting Donald Trump and Xi Jinping in Washington. The parties are now working on the practical details of this new format for cooperation.
According to Bessent, the US and China also plan to exchange expertise on the assessment and testing of the most powerful AI models before they are released. Particular attention will be paid to ensuring that humans retain control over technology.
Artificial intelligence is becoming an issue of international security
In Washington, increasing attention is being paid to the risks associated with the development of autonomous AI systems capable of performing complex tasks independently.
Among the potential threats, experts cite the uncontrolled actions of AI agents, cyber-attacks and the possible interference of automated systems in critical infrastructure.
In September, American and Chinese security experts proposed developing safeguards similar to those used to manage nuclear risks. In particular, the discussion centred on protecting nuclear command systems from AI interference and maintaining human control over military cyber operations.
At the same time, the exact list of incidents that countries will be required to report to one another has not yet been published.
Trump and Xi are set to discuss AI during their meeting in China
Further talks are scheduled for 18–19 November in Shenzhen, where the US and Chinese leaders are due to meet on the sidelines of the Asia-Pacific Economic Cooperation summit.
The meeting will be the third between Trump and Xi Jinping in 2026.
The US administration uses the term «superintelligence» to describe the direction of a new bilateral dialogue. However, this does not mean that countries have already created or control systems that surpass human intelligence in every respect.
Bessent emphasised that cooperation on security matters does not resolve strategic differences. The US and China will continue to compete for technological leadership, whilst seeking to mitigate the risks that may arise from the development of the most powerful AI models.
