As the 2026 National Two Sessions approach, it has been revealed that Zhou Hongyi, a member of the 14th National Committee of the Chinese People's Political Consultative Conference and founder, chairman, and CEO of 360, will focus his proposals on various aspects of artificial intelligence, including AI security, implementation, application, and training.
Regarding AI-enabled security, Zhou noted that companies like Anthropic have already used AI programming and vulnerability detection to solve previously intractable security challenges. He emphasized the importance of AI agents, revealing that 360 has developed tens of thousands of AI security agents capable of autonomously identifying software vulnerabilities and defending against hacker agents from other countries. These AI agents are also being used to address AI security issues. Currently, two million small and medium-sized enterprises in China are utilizing 360’s AI agents for real-time security protection.
On the topic of AI adoption in China, Zhou highlighted the country’s advantages in energy and power resources. He recommended distinguishing between two types of computing power: training compute and inference compute. While the development scale for training compute still has room for growth, the potential for inference compute is virtually limitless. Zhou urged regional development plans to prioritize inference compute.
From an industrial policy perspective, Zhou stressed the strategic importance of inference chips, advising against blindly chasing high-end training chips like those from NVIDIA. He pointed out that inference chips, along with edge-side chips for IoT and localized computing for enterprises and households, will form an extensive computing network in the future.
To help businesses and individuals quickly adopt AI, Zhou proposed the creation of an open platform for AI agents. This platform would hide the underlying infrastructure, allowing ordinary users and companies to easily build and train their own AI agents. Only by transforming computing power into specialized or industry-specific intelligence—serving as a "second brain" for individuals or internal intelligence for enterprises—can AI reach its full potential.
Additionally, Zhou called for nationwide training programs on AI agents. He explained that the development, management, and usage of AI agents differ significantly from traditional software, requiring a shift in roles for many professionals. Rather than being led by AI experts, the deployment of AI agents in enterprises should be driven by business experts. Therefore, training programs must equip business experts with the skills to develop, manage, supervise, and operate AI agents effectively.