October 3, 2026
The United States plans to propose a direct communication mechanism with China for emergencies involving artificial intelligence.
U.S. Treasury Secretary Scott Bessent said the proposal would create a process through which the two countries could notify each other if a serious AI-related incident occurs.
The comments were made in an interview with Axios and reported on October 3.
The idea reflects a growing concern among governments that increasingly capable AI systems could create incidents that cross national borders.
AI incidents are becoming an international issue
Artificial intelligence was once treated primarily as a technology-policy issue.
That is changing.
Advanced AI systems can now interact with software, generate code, conduct research and operate with increasing levels of autonomy.
An incident involving such a system could therefore potentially affect infrastructure or organizations in another country.
That creates a problem for governments.
Traditional diplomatic communication channels may not be designed for rapidly developing technical incidents involving autonomous AI systems.
Washington wants a direct communication channel
Bessent said the United States wants to establish a mechanism through which Washington and Beijing could communicate if something goes wrong with AI.
The proposal follows discussions between Bessent and Chinese Vice Premier He Lifeng about AI safety.
The objective would be practical rather than political.
If a major AI incident occurred, governments could communicate directly rather than waiting for the issue to develop through conventional diplomatic channels.
Such a mechanism could potentially cover issues involving AI systems, cybersecurity or critical infrastructure.
Open-source AI is part of the concern
According to Axios, Bessent specifically raised concerns about powerful open-source AI models developed in China.
Open models can be downloaded, modified and deployed by organizations and individuals around the world.
That creates an important distinction from proprietary AI systems.
A company can impose controls on its own model.
It is considerably harder to control a model once the underlying technology has been released and redistributed.
That is one reason governments are paying increasing attention to open AI models.
The US and China have different concerns
The two countries are approaching AI security from different perspectives.
Washington is increasingly focused on protecting critical infrastructure and reducing the risk of misuse.
China has also expressed concerns about the social and political effects of AI.
That means the countries may not agree on every aspect of AI governance.
However, they can still share an interest in preventing major incidents.
That is where an emergency communication channel could become useful.
AI safety is moving into diplomacy
The proposed mechanism is significant because it treats AI incidents in a similar way to other international risks.
Governments have established communication channels for military incidents, financial crises and other emergencies.
AI could eventually require comparable systems.
Imagine an autonomous AI system causing a major cybersecurity incident.
The first priority would be determining what happened.
The second would be preventing escalation.
The third would be communicating with other governments whose infrastructure might be affected.
A direct AI incident channel could help with that process.
Europe is watching the same problem
The issue is not limited to Washington and Beijing.
European institutions are also developing AI safety and enforcement capabilities.
The EU AI Act gives the European AI Office responsibility for enforcing rules concerning general-purpose AI models and certain high-risk systems.
The European AI Board has also discussed frontier AI capabilities, recent AI incidents and the need to strengthen Europe’s own AI capabilities and safety coordination.
That means AI incident response is gradually becoming part of government infrastructure.
Why an emergency channel could matter
The technology is developing faster than many existing international rules.
A new AI model can be released globally within hours.
An autonomous system can potentially interact with infrastructure across borders.
A vulnerability can spread internationally before governments have time to organize a response.
A direct communication mechanism would not solve those problems.
However, it could reduce the time required for governments to exchange information during a crisis.
The next step could be international AI protocols
If the U.S.–China proposal progresses, it could become an early example of a broader international approach to AI incidents.
Other countries could eventually establish similar mechanisms.
The result could be a global network for reporting and coordinating responses to major AI failures or security incidents.
That would represent another step in the evolution of AI governance.
The technology is becoming too powerful to treat as a purely domestic issue.
AI models may be developed in one country, hosted in another and used by people across the world.
That makes international communication increasingly important.
The proposed U.S.–China channel is therefore significant for a simple reason:
governments are beginning to prepare for the possibility that an AI incident could require the same kind of rapid international communication used for other global emergencies.
Laat een reactie achter