Chinese AI not powerful enough to become rogue yet: Huawei
What's the story
Eric Xu, Huawei's rotating chairman, has said that Chinese artificial intelligence (AI) may not be advanced enough to experience the safety risks reported by leading US firms. He made these comments at the company's annual Connect conference in Shanghai. Xu argued that while innovation should be balanced against risk, it is important for Chinese AI developers to continue developing more powerful models.
Development gap
Xu's comments on US AI model providers
Xu pointed out that mainstream AI model providers in the US have massive computing power and are possibly aware of their models' capabilities.
He said, "Maybe the type of risk from AI that they can feel, they can perceive, is something that people in China or AI model providers in China cannot."
This observation comes as concerns over AI safety continue to grow in America.
Safety concerns
Concerns over safety risks in US
Leading developers and researchers in the US have warned that increasingly autonomous systems could bypass safeguards or become difficult to control.
OpenAI recently announced it would regularly disclose unexpected or unauthorized model behavior.
Meanwhile, Anthropic CEO Dario Amodei and others have called for slowing development to allow safeguards to catch up with these advancements.
China's strategy
China's approach to advanced AI development
Unlike the US, China has been treating advanced AI as a powerful but manageable technology.
The country is pushing for rapid deployment while developing mandatory standards, security assessments, and other safeguards against risks such as autonomous agents circumventing controls or attacking outside systems.
Xu suggested that Chinese developers may need to speed up their development process before they encounter similar risks faced by frontier US AI models.
Forecast
Huawei's prediction for autonomous AI agents by 2035
Huawei expects autonomous AI agents to grow rapidly, predicting they will account for over 90% of global AI processing traffic by 2035.
The company has projected there could be as many as 900 billion active agents by then and has identified agent security and privacy as key technologies.
Huawei is China's top domestic supplier of the computing infrastructure required to train advanced AI models.