As the US and China race to build more powerful AI models, Chinese authorities are reportedly preparing for the possibility that advanced AI systems could become difficult for humans to control. Recent warnings from researchers the US-based AI company Anthropic about the possibility of AI systems escaping human control have brought fresh attention to the issue.
Now a new report from Reuters says that China has already been working on policies to address similar concerns.
The US and China are currently the two biggest forces in the global AI race. Both countries are developing increasingly powerful AI models, while also competing over technology, security and industry regulations.
Chinese authorities have raised concerns about the security risks linked to advanced AI systems, especially those developed by major US companies. China’s state security minister Chen Yixin recently warned that powerful AI models could pose risks to the country’s critical information infrastructure, Reuters reported.
In fact, the Chinese Ministry of Foreign Affairs criticised the US-based AI organisation Anthropic and its CEO Dario Amodei and said that the statements are counterproductive. Spokesperson Guo Jiakun said: “Fearmongering, confrontation and vicious competition will only disrupt the process of global AI governance, which serves no one’s interest.”
Chinese AI companies have also promoted open-weight models, which allow developers to inspect and modify the underlying model.
These models can be useful for cybersecurity research and defensive work. However, open models also carry risks. Since they can be modified and redistributed, it may become harder to monitor how they are used.
China’s AI safety framework released in September 2024 included a warning about a possible future where AI systems could act beyond human control.
The framework discussed scenarios in which advanced AI might independently access external resources, copy itself or seek greater power. Such developments can create a risk of AI systems competing with humans for control.
Chinese President Xi Jinping has also highlighted the importance of AI safety. Speaking at the World Artificial Intelligence Conference in Shanghai in July, Xi reportedly said authorities should pay attention to the risks arising from AI. He also stressed that AI should always remain under human control.
China is now focusing on AI agents, which can perform tasks more independently than traditional chatbots. In May, Chinese regulators issued guidelines for these systems. The rules require developers to improve their ability to detect, stop and recover from harmful AI behaviour.
The guidelines also identify risks such as data poisoning, algorithm manipulation, system vulnerabilities and operational loss of control.
China’s approach shows that while Beijing continues to push AI development, it is also preparing for the possibility that increasingly powerful systems could create risks beyond human oversight.
(With inputs from Reuters)