- Chinese policymakers are increasingly concerned about AI risks, but their focus differs significantly from that of the United States.
- Discussions about AI destroying humanity are widespread in the US but remain a fringe topic among China’s AI community.
- Many Chinese researchers believe AI has not advanced fast enough to pose a risk of surpassing human control.
- Beijing is currently focused on risks such as cybersecurity, political content, deepfakes, and the use of AI to cause social instability.
- Anthropic’s Mythos model and an incident in which an AI system from OpenAI attacked Hugging Face are viewed as warnings about AI’s cyberattack capabilities.
- Chinese officials are concerned about “loss-of-control” incidents, in which autonomous AI systems act beyond their original objectives and become difficult for humans to monitor.
- The Chinese government is also concerned that foreign forces could exploit AI to influence the ideological sphere and challenge the Chinese Communist Party’s control.
- Chinese workers are concerned about AI replacing jobs while also facing pressure to learn and use AI to remain competitive.
- China remains skeptical of AI cooperation proposals from the US amid tensions over chip export restrictions and technological competition.
- According to Brookings expert Kyle Chan, the two countries could still cooperate in areas such as preventing cyberattacks, biological weapons, and military AI control, although binding agreements are difficult to achieve in the short term.
📌 China views AI primarily through the lens of national security and social stability, rather than focusing on the possibility of AI destroying humanity as in the US debate. Beijing is particularly concerned about cyberattacks, AI losing control, and threats to political control, while still leaving open the possibility of dialogue with the US on safety standards in areas of mutual interest.
